<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>HunterDavis.com</title>
    <description>The personal blog of Hunter G. Davis (full-content feed)</description>
    <link>https://hunterdavis.com/</link>
    <language>en-US</language>
    <lastBuildDate>Wed, 11 Nov 2026 22:34:05 +0000</lastBuildDate>
    <atom:link href="https://hunterdavis.com/feed-full.xml" rel="self" type="application/rss+xml" />
    <atom:link href="https://hunterdavis.com/feed.xml" rel="alternate" type="application/rss+xml" title="Excerpt feed" />
    
      <item>
        <title>Some Switch Ports I&apos;ve Released This Past Month</title>
        <description>I’m still hard at work on decompiling and porting Pool of Radiance to the Nintendo DS. I’m also still very much hard at work on Dunking Sheep, my agentic operating system (now written in rust).</description>
        <content:encoded><![CDATA[<p>I’m still hard at work on decompiling and porting Pool of Radiance to the Nintendo DS.  I’m also still very much hard at work on Dunking Sheep, my agentic operating system (now written in rust).</p>

<p>However stoic and unaffected by trends I like to think I am, when something catches my eye I’m powerless to stop myself from diving in. With all of the decomps coming out, there’s something new every day to try, and remix, and just have fun with. So here are a couple Switch ports (and a port fix) that I’ve released this past month in the meantime.</p>

<h2 id="star-trek-elite-force">Star Trek: Elite Force</h2>

<p><img src="/content/images/2026/switch-ports/elite-force-20261110162704.jpg" alt="Elite Force on Nintendo Switch" /></p>

<p><img src="/content/images/2026/switch-ports/elite-force-20261110162826.jpg" alt="Elite Force on Nintendo Switch" /></p>

<p><img src="/content/images/2026/switch-ports/elite-force-20261110163042.jpg" alt="Elite Force on Nintendo Switch" /></p>

<p><img src="/content/images/2026/switch-ports/elite-force-20261110163020.jpg" alt="Elite Force running on Nintendo Switch" /></p>

<p>This is a fun game, and the multiplayer was already ported to switch a few years back. Finished the job for single player, you can find it here: 
https://github.com/huntergdavis/elite-force-switch</p>

<h2 id="opentrickynx">OpenTrickyNX</h2>

<p><img src="/content/images/2026/switch-ports/opentricky-20261109111531.jpg" alt="OpenTricky on Nintendo Switch" /></p>

<p>Someone else slopped a port out, but didn’t test the controls. I opened a PR to fix, working great.</p>

<p>https://github.com/huntergdavis/OpenTrickyNX</p>

<h2 id="quarantinenx">QuarantineNX</h2>

<p><img src="/content/images/2026/switch-ports/quarantinenx-20261111131748.jpg" alt="QuarantineNX on Nintendo Switch" /></p>

<p><img src="/content/images/2026/switch-ports/quarantinenx-20261111131802.jpg" alt="QuarantineNX on Nintendo Switch" /></p>

<p><img src="/content/images/2026/switch-ports/quarantinenx-20261111131812.jpg" alt="QuarantineNX running on Nintendo Switch" /></p>

<p>Quarantine 1994 was a silly fun DOS game that was kind of ahead of it’s time.  It’s a mix of crazy taxi and carmageddon.  I was having fun playing the decomp on OSX and realized it’d be an easy switch port.  Working well and it was a very quick port.</p>

<h2 id="methodology">Methodology:</h2>

<p>These are obviously heavily AI assisted.  That’s most of development now. I think the more interesting thing is how cheaply I was able to do it. I’ve got my own DS harness that handles task assignment and coordination.  My default team composition consists of one luna low writer, two terra low reviewers, and one terra low coordinator and planner.  The planner does a deep dive and cuts it up into small tasks. The writer takes one task at a time.  Code review goes through typical git/github mechanics, with branch protection and 2-reviewer sign-off. This can slow projects significantly but also catch real bugs and implementation issues early. If the project warrants it, cicd release through github with test runners. Couple of back and forth testing rounds with me. 250 million tokens, or about 9$ at API rates (this was subscription rates on a 100/mo plan, so actual cost to me 1$ or less)</p>

<p>That’s a wild number. 9$ to go from “I wish I had this game I used to play on this console I like to play on now.” to “I’m playing it!” in a few hours. Half the cost of a movie ticket (or half the cost of a candy bar if you’re subsidized by a trillion dollar corporation’s subscription plan) to share something with the world and help another gamer rekindle a lost memory. Obviously these represent the best case scenario: a straight port job where the source has already been decompiled. For more esoteric fare (i.e. pool of radiance), i’m weeks into a decompile job and i’d say i’m less than halfway there.  As model abilities continue to improve, I expect that number will drop significantly. Until then, I’m just enjoying building something. What a wild world we live in today!</p>
]]></content:encoded>
        <pubDate>Wed, 11 Nov 2026 14:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/11/11/some-switch-ports-ive-released-this-past-month.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/11/11/some-switch-ports-ive-released-this-past-month.html</guid>
      </item>
    
      <item>
        <title>Adding Modern Features to Pool of Radiance Mac Classic (1989)</title>
        <description>We are living in a tempest of a time, and my mind seeks the familiar. I picked up an 8” android eink tablet, and was pretty surprised at the lack of really good game experiences built for it’s black and white display. So I did what anyone who grew up in the 80s would do, I installed mini-VMac emulator and fired up “Pool of Radiance”</description>
        <content:encoded><![CDATA[<p>We are living in a tempest of a time, and my mind seeks the familiar. I picked up an 8” android eink tablet, and was pretty surprised at the lack of really good game experiences built for it’s black and white display. So I did what anyone who grew up in the 80s would do, I installed mini-VMac emulator and fired up “Pool of Radiance”</p>

<p>It’s brilliant, punishing, and an absolute relic of game design long past.  No map, journaling, item/weapons/armor stat lists, spell lists, levels, etc. Lots of manual lookup, with the expectation that you have graph paper handy, along with three chunky books and a code wheel.</p>

<p>“It’s a shame gold box companion wasn’t built for the Mac Classic.” I had that thought for about ten seconds.  Then I rememberd that it’s easier than ever to turn wishes to horses, so let’s ride.  A simple enough plan: fork mini-vmac to include custom views and tools for pool of radiance.</p>

<p><img src="/content/images/2026/poolrad-phlan-party-sidebar.png" alt="Live New Phlan map with party health bars, armor class and class symbols while Rolf welcomes the party in the original Mac game" width="900" /></p>

<h2 id="the-live-map">The Live Map</h2>

<p>It started with the live map, the main view that’s missing.  This works well, but I quickly realized I wanted to write notes.</p>

<p><img src="/content/images/2026/poolrad-phlan-temple-tour.png" alt="The live map tracking the party to the Temple of Tyr, with the original game's own temple artwork below it" width="900" /></p>

<h2 id="handwritten-notes-on-the-map">Handwritten Notes on the Map</h2>

<p>This was the natural extension.  You can place flags on the map, each flag when clicked opens up a scratchpad with a copy of the map you can mark up.</p>

<p><img src="/content/images/2026/poolrad-handwritten-note.png" alt="A handwritten note page with TYR sketched by hand, saved against a tile on the New Phlan map" width="900" /></p>

<p><img src="/content/images/2026/poolrad-compact-flag-note.png" alt="A handwritten flag page with compact title-row tools and twice the sketch height; the original game remains visible below" width="900" /></p>

<h2 id="reading-the-games-own-memory">Reading the Game’s Own Memory</h2>

<p>From there, I wanted to add in an easy to see tactical view (hp, status, char type).</p>

<p><img src="/content/images/2026/poolrad-combat-overview.png" alt="A read-only tactical overview of a battle, party marks filled and the rest hollow, above the game's own combat view" width="900" /></p>

<p><img src="/content/images/2026/poolrad-journal-encountered.png" alt="The notebook's encountered list filling itself in with a tavern tale and a proclamation the game named out loud" width="900" /></p>

<h2 id="offline-reference-tools">Offline Reference Tools</h2>

<p>Finally, there’s a lot of lookup tools so you don’t need to leave the game experience to enjoy it all. I added a code wheel lookup, then automated it so it happens automatically on launch, no need to play code wheel lookup or get your reading classes out.</p>

<p><img src="/content/images/2026/poolrad-companion-info.png" alt="The companion Info tab with offline spell, weapon, level and journal lookups above the running game" width="900" /></p>

<h2 id="running-it-on-e-ink">Running It on E-Ink</h2>

<p>It’s a real joy to play this on an e-ink device. It feels so analog but also kinda futuristic?  Retro futuristic.</p>

<p><img src="/content/images/2026/poolrad-eink-proclamations.png" alt="Captured on the e-ink tablet: the map filled in after a long walk, with the game posting its proclamations below" width="900" /></p>

<p><img src="/content/images/2026/poolrad-eink-tavern-tale.png" alt="Captured on the e-ink tablet: a tavern tale overheard across town, with the explored map above it" width="900" /></p>

<h2 id="getting-it">Getting It</h2>

<p>You’ll need a boot disk, apple rom, and Pool of Radiance install.  These are easy to find nowadays. Install the APK below, point to your boot disk and rom, and have fun.</p>

<p><img src="/content/images/2026/poolrad-boot-bricks.png" alt="The rebuilt boot disk starting straight into the game over a brick desktop pattern" width="900" /></p>

<p>The source, the APK releases and all of the research notes are on GitHub:
<a href="https://github.com/huntergdavis/poolrad-macmaps">huntergdavis/poolrad-macmaps</a>.</p>

<p>*Important Note -&gt; This is a proof of concept, heavily written with AI. There will be bugs, and it won’t be perfect.  This one is pretty niche, so I honestly don’t even know if anyone will check it out.  That’s OK, this one’s for me.</p>
]]></content:encoded>
        <pubDate>Tue, 15 Sep 2026 12:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/09/15/adding-modern-features-to-pool-of-radiance-mac-classic-1989.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/09/15/adding-modern-features-to-pool-of-radiance-mac-classic-1989.html</guid>
      </item>
    
      <item>
        <title>Announcing Curse of the Herder</title>
        <description>Announcing “Curse of the Herder”, a screensaver experience for the web. It’s about a pissed-off little herder learning how to cuss better.</description>
        <content:encoded><![CDATA[<p>Announcing “Curse of the Herder”, a screensaver experience for the web. It’s about a pissed-off little herder learning how to cuss better.</p>

<p><img src="/content/images/2026/curse-of-the-herder-rant.gif" alt="The cursed herder crossing the countryside, carrying sheep and muttering as he works" width="640" /></p>

<h2 id="a-cursed-working-day">A Cursed Working Day</h2>

<p>Every run is a ‘day’. Your herder starts out ignorant of the world, and of how to express themself.  They are frustrated.  They have to go collect all 60 of their sheep from the island. As they collect sheep and face frustration, their temper changes. They read books from little free libraries around the island. This is how they learn new cuss words and phrases.</p>

<p><img src="/content/images/2026/curse-of-the-herder-herder-and-dog.png" alt="The herder stands near the sheep pen with his crook, a cow, and his unhelpful sheepdog" width="900" /></p>

<p><img src="/content/images/2026/curse-of-the-herder-afternoon.png" alt="The herder carries a sheep along a country road while delivering a newly elaborate curse" width="900" /></p>

<h2 id="a-living-countryside">A Living Countryside</h2>

<p>Changing weather, seasons, other herders, little villages, small delighters. A fun little island of frustation.</p>

<p><img src="/content/images/2026/curse-of-the-herder-village-roof-sheep.png" alt="A village with cottages, a little free library, and a sheep stranded on a roof" width="900" /></p>

<p><img src="/content/images/2026/curse-of-the-herder-winter-village.png" alt="The village in winter with snow, grey trees, washing lines, and scarecrows" width="900" /></p>

<h2 id="the-end-of-the-day">The End of the Day</h2>

<p>After the last sheep is herded, the herder may retire to the hall of heroes, where their statistic and best curses are recorded.</p>

<p><img src="/content/images/2026/curse-of-the-herder-ending.png" alt="At dusk the last sheep is penned and the retiring herder's memorial stone rises beside the pen" width="900" /></p>

<h2 id="play-it">Play It</h2>

<p><strong>Live screensaver:</strong> <a href="https://hunterdavis.com/curse-of-the-herder/">hunterdavis.com/curse-of-the-herder</a></p>

<p><strong>Source:</strong> <a href="https://github.com/huntergdavis/curse-of-the-herder">github.com/huntergdavis/curse-of-the-herder</a></p>
]]></content:encoded>
        <pubDate>Wed, 09 Sep 2026 12:01:11 +0000</pubDate>
        <link>https://hunterdavis.com/2026/09/09/announcing-curse-of-the-herder.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/09/09/announcing-curse-of-the-herder.html</guid>
      </item>
    
      <item>
        <title>It happens so fast now, Johnny Castaway RetroArch released</title>
        <description>No AI was used to draft this post, these are my words fully.</description>
        <content:encoded><![CDATA[<p><em>No AI was used to draft this post, these are my words fully.</em></p>

<p>What does it mean, when your hard-fought efforts can be replaced in a moment of inspiration and a few dollars of tokens.  I find myself asking that very question today, as I watch an AI agent put the finshing touches on a port of Johnny Castaway to retroarch.</p>

<p>I mean, why not? I started on this project years ago and got distracted and discouraged and eventually distanced from it. Why wouldn’t I ask an AI to finish it up, to port Johnny to dozens (or more) of systems in one fell swoop.</p>

<p>One month after I released my magnum opus, Johnny Castaway on the PS1, which itself took every spare moment of my life for a year to release, I asked an AI to write a retroarch port. It did a great job. There’s so many systems he can run on now. It took one day.</p>

<p>A year of my life… one day. So many of us in our industry are wrestling with what that means for our profession.  I’m wrestling with what that means for my hobby. Every day I see a dozen new reverse engineering projects, ports to older systems, impossible things.  Things which used to take years and months and sweat and years of tears.</p>

<p>I’m lucky.  I spend most of my time coaching leaders, making decisions, weighing options. I know exactly what AI means to my business and my engineering departments.  Review, at a speed never imagined. Creation is no longer the bottleneck or the primary domain of humans. It’s review, curation, selection, and manual oversight after all the AI guardrails and systems.  Always a human reviewing. That’s medical software in 2026, and I don’t think that’ll change next year. At home though?  In entertainment?  In games?  In mods and hacks and custom firmware and reverse engineering?  The human is so often becoming just the spark at the beginning. And I find myself thinking of that.  One year… one day. How much what I expect from my computing environment has changed even day to day. I’m still in the terminal, but i’m in a terminal on some far future sci-fi fantasy novel. Every day, multiplexed and dunked 10x.</p>

<p>Do I mourn what’s lost or celebrate what’s gained? Even setting aside the societal cost, it’s made for a turbulence and a negative shift in my industry. I’m on optimist though, and I believe we can lead our way through it with empathy and understanding.</p>

<p>That said, here’s Johnny Castawy ported to retroarch.  I’ve kept some of my personal improvements (closed captions, scene descriptions, chapter select, extra holidays.) This should run on a lot more systems now, here’s the full list (library and package releases):</p>

<ul>
  <li>
    <p>Linux
    - x86_64
    - x86
    - ARM64
    - ARMv7/VFP
    - ARMv7/NEON</p>

    <ul>
      <li>Windows
        <ul>
          <li>x64</li>
          <li>x86</li>
        </ul>
      </li>
      <li>Android
        <ul>
          <li>ARM64</li>
          <li>ARMv7</li>
          <li>x86_64</li>
          <li>x86</li>
        </ul>
      </li>
      <li>Apple
        <ul>
          <li>macOS (Intel and Apple Silicon)</li>
          <li>iPhone and iPad</li>
          <li>Apple TV</li>
          <li>iOS and tvOS simulators</li>
        </ul>
      </li>
      <li>Web
        <ul>
          <li>Emscripten/RetroArch Web</li>
        </ul>
      </li>
      <li>Sony
        <ul>
          <li>PlayStation Portable</li>
          <li>PlayStation Vita and Vita TV</li>
          <li>PlayStation 2</li>
        </ul>
      </li>
      <li>Nintendo
        <ul>
          <li>Nintendo Switch</li>
          <li>Nintendo 3DS</li>
          <li>GameCube</li>
          <li>Wii</li>
          <li>Wii U</li>
        </ul>
      </li>
    </ul>
  </li>
</ul>

<p>Direct installable packages are available for :</p>

<ul>
  <li>PSP: EBOOT.PBP with a Memory Stick directory layout</li>
  <li>Switch: .nro</li>
  <li>Nintendo 3DS: .3dsx and .cia</li>
  <li>GameCube and Wii: .dol</li>
  <li>Wii U: .rpx</li>
  <li>PlayStation Vita: .vpk</li>
  <li>PlayStation 2: .elf</li>
</ul>

<p>Here’s the GitHub project: <a href="https://github.com/huntergdavis/johnny-castaway-retroarch">github.com/huntergdavis/johnny-castaway-retroarch</a></p>

<p>Here’s the project site: <a href="https://hunterdavis.com/johnny-castaway-retroarch/">hunterdavis.com/johnny-castaway-retroarch</a></p>

<p>Here’s the <a href="https://hunterdavis.com/johnny-castaway-retroarch/play/">live Web player</a> (RetroArch version, which makes this my fourth JC Web player).</p>
]]></content:encoded>
        <pubDate>Fri, 28 Aug 2026 20:57:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/08/28/it-happens-so-fast-now-johnny-castaway-retroarch-released.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/08/28/it-happens-so-fast-now-johnny-castaway-retroarch-released.html</guid>
      </item>
    
      <item>
        <title>Steam on Android is Workable, Performant, and Brittle.</title>
        <description>A few weeks ago I wrote about getting the full ARM64 Steam client running on an unrooted Android tablet. At the time it was a proof of concept. Steam opened, authenticated, downloaded games, and launched a few of them through Proton 11 ARM64, FEX, DXVK, and Turnip. That felt like a good start to share.</description>
        <content:encoded><![CDATA[<p>A few weeks ago I wrote about getting the <a href="/2026/08/09/full-steam-on-termux-because-why-not.html">full ARM64 Steam client running on an unrooted Android tablet</a>.  At the time it was a proof of concept. Steam opened, authenticated, downloaded games, and launched a few of them through Proton 11 ARM64, FEX, DXVK, and Turnip. That felt like a good start to share.</p>

<p>The good news is that Steam on Android is absolutely workable. It can also be surprisingly performant. The bad news is that the whole stack sits on pillars of sand, slight mismatches between expectations and abilities. Linux versus Android. glibc versus Bionic. One Android UID versus another. Internal F2FS versus an exFAT SD card behind Android’s FUSE layer. X11 versus an Android Surface. A native ARM64 Steam client versus x86 Windows games. Every one of those boundaries can work, often on a per-game basis, then a new game changes expectations and the whole thing becomes brittle again.</p>

<p>This is what I had to do to make it work well, and why I now recommend <a href="https://github.com/utkarshdalal/GameNative">GameNative</a> for almost everyone who just wants to play games.</p>

<h2 id="the-stack-that-worked">The Stack That Worked</h2>

<p>The test device is still my unrooted Galaxy Tab S8+, a Snapdragon 8 Gen 1 tablet with an Adreno 730, running stock Android. Fully stock and unrooted.</p>

<p>The working path ended up looking like this:</p>

<div class="language-text highlighter-rouge"><div class="highlight"><pre class="highlight"><code>Android / Samsung kernel
  -&gt; Termux application sandbox
    -&gt; shared-UID Termux:X11
    -&gt; native ARM64 Steam in a glibc userspace
      -&gt; software-rendered Chromium for the Steam interface
      -&gt; Steam Runtime 4 ARM64
      -&gt; Proton 11 ARM64
      -&gt; FEX for the x86 Windows game
      -&gt; DXVK
      -&gt; private Mesa Turnip
      -&gt; Adreno 730
</code></pre></div></div>

<p>Steam itself is ARM64 here. FEX enters the picture later, when Proton launches an x86 or x86-64 Windows game. The first half of this project was not about CPU emulation at all. It was about convincing a normal Linux program that Android behaves enough like the Linux system it expects.</p>

<figure>
  <a href="/content/images/2026/steam-termux-superflight-gameplay.png"><img src="/content/images/2026/steam-termux-superflight-gameplay-small.png" alt="Superflight running fullscreen through Steam, Proton, FEX, DXVK, and Turnip on Android" width="640" height="363" /></a>
  <figcaption>The original proof of concept: Superflight running through the complete Steam-to-Adreno stack on the tablet.</figcaption>
</figure>

<h2 id="android-has-a-linux-kernel-but-it-is-not-a-glibc-linux-desktop">Android Has a Linux Kernel, but It Is Not a glibc Linux Desktop</h2>

<p>Android’s native C library is Bionic. Conventional Linux Steam expects glibc, conventional paths, conventional process behavior, and a set of old Unix facilities which Android applications do not necessarily get.</p>

<figure>
  <a href="/content/images/2026/steam-arm64-loading-termux.webp"><img src="/content/images/2026/steam-arm64-loading-termux-small.webp" alt="ARM64 Steam loading user data in front of a Termux terminal on Android" width="800" height="250" /></a>
  <figcaption>ARM64 Steam loading inside the Android-hosted Linux desktop. Getting this window was easy compared with reproducing everything behind it.</figcaption>
</figure>

<p>The first working version put Steam inside a Debian PRoot. That gave it the filesystem and glibc layout it wanted, but stock PRoot was not enough. I ended up implementing narrowly scoped patches for:</p>

<ul>
  <li>robust-list calls blocked by Android’s seccomp policy;</li>
  <li>System V semaphores, including the wakeups Steam expects when another process changes a semaphore;</li>
  <li>Pressure Vessel’s <code class="language-plaintext highlighter-rouge">.l2s</code> metadata, hard-link representation, bind mounts, pivot sequence, and shared <code class="language-plaintext highlighter-rouge">/tmp</code> behavior;</li>
  <li>path translation and mountinfo escaping;</li>
  <li>the private <code class="language-plaintext highlighter-rouge">/proc/net</code> route view Wine uses for Windows networking APIs; and</li>
  <li>the traditional Steam SDK paths and ARM64 Runtime/Proton registration which the new client still expected to find.</li>
</ul>

<p>I had a real, authenticated Steam client and real games. It also meant every interesting filesystem operation passed through a <code class="language-plaintext highlighter-rouge">ptrace</code>-based compatibility layer, and it was slow as hell.</p>

<p>That is fine for a shell command. It is much less fine when Steam scans thousands of tiny Runtime and Proton files and takes 10 minutes to load. Had to find another way!</p>

<h2 id="the-glibc-wrapper">The glibc Wrapper</h2>

<p>The <a href="https://github.com/huntergdavis/steamclienttermux/blob/main/docs/PERFORMANCE.md">metadata benchmark</a> gave me the next target. This was a comparative test, and when compared with native runtime behavior, enumerating the same 5,601 Proton files took 0.129 seconds natively and 4.231 seconds through the production PRoot path. Multiply this by every system and cold boots We’re taking 20x longer than they should due to Proot overhead.</p>

<p>I tried the simplest optimizations first. Removing unnecessary PRoot extensions cut the test to about 2.10 seconds. A guarded fast path for a narrow class of <code class="language-plaintext highlighter-rouge">fstatat</code> calls reached about 1.76 seconds. That was a useful 2.4x improvement over the original route, but it was still about 14x slower than native filesystem access.</p>

<p>So I built <a href="https://github.com/huntergdavis/termux-glibc-compat">termux-glibc-compat</a>: a focused compatibility layer for running glibc applications directly inside the Termux application sandbox without sending every syscall through PRoot.</p>

<p>Files, sockets, futexes, graphics, and ordinary process execution stay on native kernel paths. A same-UID Bionic broker supplies the missing System V semaphore operations over an authenticated Unix socket. A patched Termux glibc calls that broker at the stable libc boundary. A small robust-list shim satisfies the userspace contract Steam checks, while being explicit that Android’s kernel still does not know about the synthetic list.</p>

<p>Then came the small, annoying edges which make up a real compatibility layer:</p>

<ul>
  <li>an execution shim wraps AArch64 Linux children with the selected glibc loader without copying or patching them;</li>
  <li>exact <code class="language-plaintext highlighter-rouge">/bin/sh</code> and <code class="language-plaintext highlighter-rouge">/usr/bin/sh</code> redirection prevents Steam from accidentally executing Android’s Bionic shell as if it were a glibc ELF;</li>
  <li>read-only <code class="language-plaintext highlighter-rouge">/proc/net</code> and <code class="language-plaintext highlighter-rouge">/proc/stat</code> shadows restore the bits Wine and CPU-topology detection need;</li>
  <li>Android’s fatal <code class="language-plaintext highlighter-rouge">SIGSYS</code> for <code class="language-plaintext highlighter-rouge">userfaultfd</code> is normalized to the <code class="language-plaintext highlighter-rouge">ENOSYS</code> failure Proton already understands;</li>
  <li>removable-storage <code class="language-plaintext highlighter-rouge">flock</code> failures can fall back to record locks for the measured single-client Steam path; and</li>
  <li>executable mappings from <code class="language-plaintext highlighter-rouge">noexec</code> storage can be replaced with byte-identical anonymous mappings when Wine changes page protections.</li>
</ul>

<p>This moved the authenticated Steam client and its Chromium helpers out of PRoot. The <a href="https://github.com/huntergdavis/termux-glibc-compat/blob/main/docs/PERFORMANCE.md#2026-08-17-native-steam-and-controlled-game-measurements">comparable Tomb Raider launch</a> went from 407.236 seconds between the Runtime request and the first game window to 58.256 seconds. 7x speed improvement and usable.</p>

<figure>
  <a href="/content/images/2026/tombraider-direct-glibc-no-proot.webp"><img src="/content/images/2026/tombraider-direct-glibc-no-proot-small.webp" alt="Tomb Raider terms screen reached through the direct glibc compatibility path without PRoot" width="800" height="501" /></a>
  <figcaption>The first Tomb Raider window from the direct glibc host path, with the remaining PRoot launch boundary removed.</figcaption>
</figure>

<h2 id="android-schedules-applications-not-just-processes">Android Schedules Applications, Not Just Processes</h2>

<p>The most frustrating performance discovery had nothing to do with FEX, DXVK, or Vulkan. It was Android foreground ownership.</p>

<p>Termux runs the Linux processes. Termux:X11 owns the visible Android activity. With the standalone Termux:X11 APK, those can be different Android UIDs. When the small Termux window disappeared behind the fullscreen X11 activity, Android decided the Termux UID was a background application. Tomb Raider was pushed into <code class="language-plaintext highlighter-rouge">/background</code> and <code class="language-plaintext highlighter-rouge">/cpuset/moderate</code>, with only a few slower CPU cores available.</p>

<p>At the same 1280x720, Low, V-Sync-off settings, the <a href="https://github.com/huntergdavis/steamclienttermux/blob/main/docs/TOMB_RAIDER_BENCHMARK.md#full-screen-termuxx11-usability-ab">standalone-UID fullscreen run</a> averaged 5.4 FPS.</p>

<p>The official Termux:X11 project publishes a <a href="https://github.com/termux/termux-x11#avoiding-slowdowns"><code class="language-plaintext highlighter-rouge">sharedUid</code> build specifically to avoid this slowdown</a>. After installing the matching build, Termux, Termux:X11, and Termux:API all reported the same Android UID. Now the visible X11 activity kept the Linux game tree in Android’s <code class="language-plaintext highlighter-rouge">/top-app</code> groups. The next fullscreen run averaged 28.5 FPS.</p>

<figure>
  <a href="/content/images/2026/tombraider-shared-uid-benchmark.webp"><img src="/content/images/2026/tombraider-shared-uid-benchmark-small.webp" alt="Tomb Raider benchmark result showing 17.4 minimum, 36.3 maximum, and 28.5 average FPS" width="800" height="450" /></a>
  <figcaption>The captured shared-UID result: 17.4 minimum, 36.3 maximum, and 28.5 average FPS.</figcaption>
</figure>

<p>Processor affinity accounted for a 6x slowdown. Same game. Same settings.</p>

<p>This also explains some wonderfully confusing debugging sessions where a game performed well with a tiny Termux window floating over it and collapsed as soon as I made the game properly fullscreen. The little window was not helping graphics. It was keeping the correct UID in the foreground.</p>

<p>Once that was fixed, I could tune the actual workload. The useful production layout gave Steam’s web helpers CPU 0, kept Termux:X11 on CPUs 0-3, allowed the game on CPUs 1-7, and kept one continuously busy networking thread on CPU 1. I tested FEX profiles, thermal starting conditions, display refresh, process affinity, CEF suspension, and several ideas which sounded good at the time but measured worse.</p>

<p>Counterintuitively, changing the X11 surface from 119.92 Hz to 59.97 Hz improved the panel-native Tomb Raider average from 23.400 to 25.167 FPS. Moving the hot game path out of its remaining PRoot boundary later reached about 30.6 FPS with the safer FEX profile. The more aggressive profile was not meaningfully faster, so the boring safe configuration won.</p>

<h2 id="the-bionic-vulkan-bridge-rabbit-hole">The Bionic Vulkan Bridge Rabbit Hole</h2>

<p>The production graphics path was already respectable:</p>

<div class="language-text highlighter-rouge"><div class="highlight"><pre class="highlight"><code>DXVK -&gt; glibc Vulkan loader -&gt; private Mesa Turnip -&gt; Adreno
</code></pre></div></div>

<p>Steam’s Chromium UI did not like GPU compositing through Termux:X11, so the client UI stayed on software rendering while games kept hardware Vulkan. Hacky division of labor, but effective.</p>

<p>Then I started wondering whether a glibc game could use Android’s Bionic Vulkan loader and vendor driver directly. This became <a href="https://github.com/huntergdavis/bionic-vulkan-bridge">bionic-vulkan-bridge</a>, a rabbit hole I lived in for a week.</p>

<p>The simple diagram looked reasonable:</p>

<div class="language-text highlighter-rouge"><div class="highlight"><pre class="highlight"><code>glibc game / DXVK
  -&gt; Vulkan ICD bridge
  -&gt; Bionic service
  -&gt; Android Vulkan loader
  -&gt; Adreno driver
  -&gt; Android Activity surface
</code></pre></div></div>

<p>The implementation was anything but simple. Vulkan is a large, chatty, handle-heavy API. Pointers and C structures cannot simply cross a glibc/Bionic process boundary. Android also prevents two arbitrary app UIDs from passing some kinds of sockets and file descriptors directly.</p>

<p>I ended up building typed proxy handles, versioned wire structures, command batching, authenticated loopback control, shared-memory rings, Binder setup callbacks, same-UID <code class="language-plaintext highlighter-rouge">SCM_RIGHTS</code> relays, external-memory images, <code class="language-plaintext highlighter-rouge">SYNC_FD</code> ordering, a virtual swapchain, an Android fullscreen Activity, persistent image ownership, descriptor journals, and enough generated dispatch to get real DXVK work through the bridge.</p>

<p>It worked, but was slow as hell. The bridge enumerated the real GPU, submitted commands, rendered test triangles, shared images across the Android boundary, produced real Tomb Raider frames, and eventually completed the game’s built-in benchmark. It was also dramatically slower than the ordinary Turnip path. The <a href="https://github.com/huntergdavis/bionic-vulkan-bridge/blob/main/docs/evidence/e134-tombraider-complete-benchmark-tablet.json">background-scheduled run</a> averaged 2.2 FPS. <a href="https://github.com/huntergdavis/bionic-vulkan-bridge/blob/main/docs/evidence/e135-tombraider-top-app-speedup-tablet.json">Keeping the Termux side in <code class="language-plaintext highlighter-rouge">/top-app</code></a> raised that to 7.0 FPS, proving both that the scheduling issue mattered and that the bridge still had a great deal of synchronization and submission overhead.</p>

<p>That experiment was valuable. It proved what the boundaries actually are. It also proved that turning Vulkan into a remote procedure call is not a shortcut to performance. A call which costs almost nothing inside one process can become disastrous when a game makes thousands of them across a serialization and synchronization boundary.</p>

<figure>
  <a href="/content/images/2026/bionic-vulkan-bridge-triangle-and-tombraider.webp"><img src="/content/images/2026/bionic-vulkan-bridge-triangle-and-tombraider-small.webp" alt="Bionic Vulkan bridge rendering a test triangle and then a real Tomb Raider frame through an Android X11 activity" width="800" height="251" /></a>
  <figcaption>The bridge progression in two real captures: an Android-backed Vulkan triangle with an unresponsive-activity warning, then a real Tomb Raider frame through the same boundary.</figcaption>
</figure>

<p>The best optimization was to avoid needing that bridge in the first place.</p>

<h2 id="an-sd-card-is-not-just-a-slower-directory">An SD Card Is Not Just a Slower Directory</h2>

<p>Large PC games do not belong on a tablet’s small internal filesystem, so the microSD card became another compatibility project.</p>

<p>Android exposed my exFAT card through FUSE with <code class="language-plaintext highlighter-rouge">noexec</code> and no symlink support. <code class="language-plaintext highlighter-rouge">flock</code> worked on internal F2FS and returned <code class="language-plaintext highlighter-rouge">ENOSYS</code> on the card. Steam interpreted that as a failure to allocate a file reader, a failure to write patch state, or corrupt reusable depot chunks depending on which code path hit it. Proton prefixes also need Linux filesystem behavior which the card cannot provide.</p>

<p>The answer was a <a href="https://github.com/huntergdavis/steamclienttermux/blob/main/docs/TECHNICAL_LOG.md#2026-08-10-removable-windows-game-library">split Steam library</a>:</p>

<div class="language-text highlighter-rouge"><div class="highlight"><pre class="highlight"><code>internal F2FS
  library metadata and appmanifests
  compatdata / Proton prefixes
  active download state
  runtimes, Proton, FEX, and native executables

microSD
  steamapps/common game payloads
  large per-game staging trees
</code></pre></div></div>

<p>The launcher assembles those into one Steam-visible directory with exact symlinks for the native client and ordered nested binds at the PRoot game boundary. Small, lock-sensitive, executable, and metadata-heavy files stay internal. Large Windows game payloads live on the card.</p>

<p>That arrangement made Kingsway survive a restart and run directly from microSD. GTA IV exposed the next problem: committing almost 24 GB across filesystems one translated metadata operation at a time was painfully slow. The staging tool therefore copied and hash-verified the tree onto the card first, exposed that same-device staging directory at Steam’s expected download path, and completed the final commit with native same-filesystem renames.</p>

<figure>
  <a href="/content/images/2026/steam-sd-card-kingsway-gtaiv.webp"><img src="/content/images/2026/steam-sd-card-kingsway-gtaiv-small.webp" alt="Kingsway and Grand Theft Auto IV launched from Steam game payloads stored on the Android microSD card" width="800" height="227" /></a>
  <figcaption>Kingsway survived a restart and ran from microSD; GTA IV made the cost of committing a much larger cross-filesystem payload impossible to ignore.</figcaption>
</figure>

<p>It works. It is also a lot of machinery to preserve the illusion that Steam is talking to a normal Linux filesystem.</p>

<h2 id="then-i-read-the-gamenative-source">Then I Read the GameNative Source</h2>

<p>I had previously grouped GameNative with the other convenient Android PC-game launchers and kept it at arm’s length because I did not want to hand my Steam credentials to a black box. Then I found out that <a href="https://github.com/utkarshdalal/GameNative/blob/master/README.md">GameNative is GPL-3.0 open source</a>.</p>

<p>Reading the source was one lightbulb moment after another. GameNative had already solved, or more often architecturally avoided, several of our hardest problems.</p>

<figure>
  <a href="/content/images/2026/gamenative-library-ui.webp"><img src="/content/images/2026/gamenative-library-ui-small.webp" alt="GameNative Android library interface showing Steam, GOG, Epic, Amazon, and custom game tabs" width="800" height="450" /></a>
  <figcaption>GameNative's Android-native library UI, captured from the <a href="https://github.com/utkarshdalal/GameNative#readme">project's official demo</a>. This is the architectural inversion that makes the rest of its approach possible.</figcaption>
</figure>

<p>Its <a href="https://github.com/utkarshdalal/GameNative/blob/master/app/src/main/java/com/winlator/xenvironment/components/BionicProgramLauncherComponent.java">Bionic program launcher</a> runs Wine, FEX or Box64, input shims, Android shared-memory support, graphics setup, networking, and the Steam bridge inside an Android-native environment. Its X server, renderer, surface, input system, and visible Activity belong to the application rather than being spread across a conventional glibc distro and a separate display APK. There is no reason to remote every Vulkan call across the bridge I had just spent days building.</p>

<p>GameNative also does not need to preserve the entire conventional Linux Steam desktop client architecture. It has its own <a href="https://github.com/utkarshdalal/GameNative/blob/master/app/src/main/java/app/gamenative/service/SteamService.kt">Steam service and depot downloader</a>, then provides the Steamworks pieces a launched game needs. Its experimental Bionic Steam path wires the matching <code class="language-plaintext highlighter-rouge">lsteamclient</code> pieces to a native Android <code class="language-plaintext highlighter-rouge">libsteamclient.so</code> and takes care of the per-Proton-version assets and application identity.</p>

<p>The storage policy is familiar too. Game installs can live on external volumes while containers and sensitive state remain internal. The current <a href="https://github.com/utkarshdalal/GameNative/blob/master/app/src/main/java/app/gamenative/utils/StorageUtils.kt#L126-L150">storage code</a> even moves new external installs from <code class="language-plaintext highlighter-rouge">Android/data</code> to a public <code class="language-plaintext highlighter-rouge">GameNative</code> directory because Android disables useful FUSE caching under the app-specific tree, making metadata operations catastrophically slow. There is a real <a href="https://github.com/utkarshdalal/GameNative/blob/master/app/src/main/java/app/gamenative/utils/ContainerStorageManager.kt">storage manager</a> for moving individual games instead of a pile of hand-assembled Steam library binds.</p>

<p>This does not mean GameNative is perfect. Its own <a href="https://github.com/utkarshdalal/GameNative/blob/master/ROADMAP.md">roadmap</a> still lists external launcher reliability, Steam Input, external storage, device compatibility, performance, and thermals as active work. Windows gaming on Android remains a pile of compatibility projects no matter how nice the interface is.</p>

<p>But it is the right architecture for most people. It starts as an Android application and brings in the PC-game compatibility pieces it needs. My Termux project starts with conventional Linux Steam and spends its time repairing every place where that assumption collides with Android.</p>

<h2 id="my-recommendation">My Recommendation</h2>

<p>If you want to learn how far unrooted Android can be pushed, need a conventional authenticated ARM64 Linux Steam client, or enjoy turning syscall traces into weekend plans, the <a href="https://github.com/huntergdavis/steamclienttermux">Steam ARM64 on Termux/X11 project</a> is real and repeatable. I am glad I built it. The <a href="https://github.com/huntergdavis/termux-glibc-compat">glibc compatibility layer</a> is useful beyond Steam, and the Vulkan bridge taught me more about Android graphics and process boundaries than any sane project would have.</p>

<p>If you want to play your PC games on Android, use <a href="https://gamenative.app/">GameNative</a>.</p>

<p>That is not an admission that Steam on Android failed. It is the opposite. Steam on Android is workable. With the right storage layout, scheduling, native glibc host, and graphics path, it is performant. The amount of work required to keep all of those layers aligned is exactly what makes it brittle.</p>

<p>GameNative packages together solutions for so many problems you don’t realize you’re facing. For almost everyone, that is the solution I recommend.</p>
]]></content:encoded>
        <pubDate>Wed, 26 Aug 2026 08:30:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/08/26/steam-on-android-is-workable-performant-and-brittle.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/08/26/steam-on-android-is-workable-performant-and-brittle.html</guid>
      </item>
    
      <item>
        <title>Full GPU Accelerated ARM64 Steam on Unrooted Termux, Because Why Not?</title>
        <description>I’ve been having so much fun running full steam games on my switch lite! Just this weekend I got Psychnauts running well (with no overclocks!) These can be good times even if they are complicated times. Sometimes, you just need to know something is possible before it becomes easy for everyone. That’s what I’m hoping for here. I’ve got a proof of concept that’s working well enough to share, but I’m sure it could be much better done in a bespoke manner by a developer with more time. Full graphics-accelerated steam and steam games, inside a termux linux session , running on an unrooted stock Galaxy Tab S8+ (SM-X808U, Snapdragon 8 Gen 1 / Adreno 730).</description>
        <content:encoded><![CDATA[<p>I’ve been having so much fun running full steam games on my switch lite! Just this weekend I got Psychnauts running well (with no overclocks!) These can be good times even if they are complicated times. Sometimes, you just need to know something is possible before it becomes easy for everyone. That’s what I’m hoping for here.  I’ve got a proof of concept that’s working well enough to share, but I’m sure it could be much better done in a bespoke manner by a developer with more time. Full graphics-accelerated steam and steam games, inside a termux linux session , running on an unrooted stock Galaxy Tab S8+ (SM-X808U, Snapdragon 8 Gen 1 / Adreno 730).</p>

<p>Here’s how it came together. Obviously I’ve been enjoying how easy it’s become to run steam games on the switch lite. Gamehub and GameNative and Winlator are all great pieces of software, but I’m never going to really feel comfortable putting my steam auth information into a 3rd party tool. There will always be some spectre of doubt and the fear that my account could somehow be compromised. So the desire is there, latent, waiting for the situation.</p>

<p>Then it hits me.  I’ve got a tablet that I use for programming while I’m on the road.  It’s got 5g internet built in, and I got a sweet deal on a payment plan from T-Mobile.  Cheap and quite powerful. I have Termux installed with Termux-X11 and graphics acceleration working. I run herdr, coding harnesses, compilers, a full KDE desktop (sans root). It works better than you’d expect, franky most folks would think I’m just running a kde distro on a small tablet computer, if not for the T-Mobile chime on boot. Why then, should I have to swap out to some third party to play my steam games? ARM64 Steam should work (tldr; it does)</p>

<p>So I fire up an agent, tell it to start monitoring syscalls, and I give it a shot. One problem at a time.  This is missing, that is crashing. Before too long the steam interface comes up.  Then EA. Then some games. A little progress, but progress. A small proot patch, some launch scripts, nothing wacky.</p>

<p><a href="/content/images/2026/steam-termux-superflight-gameplay.png"><img src="/content/images/2026/steam-termux-superflight-gameplay-small.png" alt="View the full-size Superflight game running through Steam on Termux" width="640" height="363" /></a></p>

<p>Yes, I used a coding assistant LLM to perform rote tasks, generate code, and run tests overnight.  No, I don’t think that means the code is poisoned by the well of AI. I just wanted to show that it was possible, and not far out of reach. I really do think Valve will release something like this but 100 times better on their own, eventually. In the meantime here’s the code:</p>

<p><a href="https://github.com/huntergdavis/steamclienttermux">Steam ARM64 on Termux/X11</a></p>

<p>Is this better than gamenative, winlator, gamehub, etc?  Probably not.  Is it easier to setup than those, definitely not.  Is it a step in the right direction though…? I think so.</p>
]]></content:encoded>
        <pubDate>Sun, 09 Aug 2026 18:25:16 +0000</pubDate>
        <link>https://hunterdavis.com/2026/08/09/full-steam-on-termux-because-why-not.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/08/09/full-steam-on-termux-because-why-not.html</guid>
      </item>
    
      <item>
        <title>Blur, Brutal Legend, C&amp;C Red Alert 3: Linux Does What Nintendon&apos;t</title>
        <description>The most valuable thing you can do for the longevity of a device is install linux on it. I’m not just talking personal longevity, I’m also talking about the weird and wonderful ways you can keep a platform alive when you have an open target.</description>
        <content:encoded><![CDATA[<p>The most valuable thing you can do for the longevity of a device is install linux on it. I’m not just talking personal longevity, I’m also talking about the weird and wonderful ways you can keep a platform alive when you have an open target.</p>

<p>That brings me to my switch lite.  Probably my favorite console of all time (and I’ve had them all.) I love the switch, but when the switch lite came out I was lucky enough to get the official nintendo hard-case. It’s indestructable, you can throw it into a backpack or suitcase destined to be smashed and beat the hell up with no worry.  My many plane rides were completely transformed.</p>

<p>Fast forward to July 2026.  I’ve so many handhelds, an actual Steam Deck, phones, tablets, gaming pcs, yada yada.  And I spy it.  A link shared in the comments thread on a github repo. They’ve gottten the ARM release of steam running on Tegra (Switch) Linux. My mind is set ablaze!  Though the hardware is basically 2011 tablet/phone hardware, that’s still powerful enough to run many generations of games.  Games that never came out on the switch. Games that will never be re-released or ported anyhwere near a Nintendo system. PC games that won’t run in retroarch, real deal AAA games.</p>

<p>One week later and I’ve got three of my absolute favorites running quite well. These are games that everyone should play, and to have them on my Switch Lite has been a delight. Fills my heart with joy.  Curious?  Here’s the setup for each.</p>

<h2 id="blur-steam-appid-42640">Blur (Steam appid 42640)</h2>

<p><a href="/content/images/2026/switchdeck-steam-blur-1.png"><img src="/content/images/2026/switchdeck-steam-blur-1-small.png" alt="View the full-size Blur gameplay screenshot showing cars colliding in a burst of power-up effects" width="640" height="360" /></a></p>

<ol>
  <li>Replace <code class="language-plaintext highlighter-rouge">Blur.exe</code> with a no-CD / no-GFWL exe (the stock GFWL wrapper hangs under box64).</li>
  <li>Launch via Proton — the no-CD exe has no Steamworks DRM, so it can run without Steam.</li>
  <li>
    <p>Float the Joy-Con off X11 (stops stick→cursor and buttons→Escape):</p>

    <div class="language-bash highlighter-rouge"><div class="highlight"><pre class="highlight"><code>xinput list                    <span class="c"># find the two "Nintendo Switch Lite Gamepad" ids</span>
xinput float &lt;pointer-id&gt;      <span class="c"># stick no longer moves the cursor</span>
xinput float &lt;keyboard-id&gt;     <span class="c"># buttons no longer send Escape</span>
<span class="c"># on exit, restore:</span>
xinput reattach &lt;pointer-id&gt;  <span class="s2">"Virtual core pointer"</span>
xinput reattach &lt;keyboard-id&gt; <span class="s2">"Virtual core keyboard"</span>
</code></pre></div>    </div>
  </li>
  <li>
    <p>box64 flags + performance governor (it’s CPU-bound; GPU is mostly idle):</p>

    <div class="language-bash highlighter-rouge"><div class="highlight"><pre class="highlight"><code><span class="c"># Steam launch options / env:</span>
<span class="nv">BOX64_DYNAREC_BIGBLOCK</span><span class="o">=</span>3 <span class="nv">BOX64_DYNAREC_STRONGMEM</span><span class="o">=</span>0 <span class="nv">BOX64_DYNAREC_SAFEFLAGS</span><span class="o">=</span>0 %command%

<span class="c"># lock CPU cores to max clock:</span>
<span class="nb">echo </span>performance | <span class="nb">sudo tee</span> /sys/devices/system/cpu/cpu<span class="k">*</span>/cpufreq/scaling_governor
</code></pre></div>    </div>
  </li>
</ol>

<h2 id="brütal-legend-steam-appid-225260">Brütal Legend (Steam appid 225260)</h2>

<div class="grid">
  <div class="unit one-third"><a href="/content/images/2026/switchdeck-steam-brutal-legend-1.png"><img src="/content/images/2026/switchdeck-steam-brutal-legend-1-small.png" alt="View the full-size Brütal Legend opening cutscene with Jack Black browsing records" width="320" height="180" /></a></div>
  <div class="unit one-third"><a href="/content/images/2026/switchdeck-steam-brutal-legend-2.png"><img src="/content/images/2026/switchdeck-steam-brutal-legend-2-small.png" alt="View the full-size Brütal Legend title screen presented as a record sleeve" width="320" height="180" /></a></div>
  <div class="unit one-third"><a href="/content/images/2026/switchdeck-steam-brutal-legend-3.png"><img src="/content/images/2026/switchdeck-steam-brutal-legend-3-small.png" alt="View the full-size Brütal Legend gameplay screenshot showing the Deuce driving through the metal landscape" width="320" height="180" /></a></div>
</div>

<ol>
  <li>Use <strong>GE-Proton 10-34</strong>. Runs through Steam (it uses Steamworks DRM, so Steam stays running).</li>
  <li>Add your controller’s SDL GUID to the game’s per-user <code class="language-plaintext highlighter-rouge">SDLGamepad.config</code> (its stock list is too old to recognize modern pads).</li>
  <li>Float the Joy-Con off X11 (same commands as Blur step 3), and set the game’s <strong>Steam Input to Disabled</strong>.</li>
  <li>Add <code class="language-plaintext highlighter-rouge">SD_GAMEMODE=1</code> to the launch options to unload Steam’s UI (~600 MB) while keeping the client alive for DRM.</li>
</ol>

<h2 id="command--conquer-red-alert-3-steam-appid-17480">Command &amp; Conquer: Red Alert 3 (Steam appid 17480)</h2>

<div class="grid">
  <div class="unit half"><a href="/content/images/2026/ccra3-1.png"><img src="/content/images/2026/ccra3-1-small.png" alt="View the full-size Command &amp; Conquer: Red Alert 3 Soviet campaign briefing" width="480" height="270" /></a></div>
  <div class="unit half"><a href="/content/images/2026/ccra3-2.png"><img src="/content/images/2026/ccra3-2-small.png" alt="View the full-size Command &amp; Conquer: Red Alert 3 gameplay showing a unit moving beside a coastal road" width="480" height="270" /></a></div>
</div>

<p>This one just needed some time to bake.  After apply the above generic fixes, it still didn’t seem like it was loading.  Turns out the first launch takes like 15 minutes for whatever reason, subsequent launches are in the 30 second range.  Finally I have Tim Curry properly on my Switch !!</p>

<h2 id="one-gotcha-for-controller-games">One Gotcha for Controller Games:</h2>

<p>Switchroot’s Joy-Con driver exposes the pad as an <strong>X11 mouse + keyboard</strong> on top of the real gamepad.  Native-gamepad games get double input (stick moves the cursor, buttons fire Escape and quit). Use <code class="language-plaintext highlighter-rouge">xinput float</code> to disable as noted above.</p>
]]></content:encoded>
        <pubDate>Fri, 31 Jul 2026 10:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/07/31/blur-brutal-legend-ccra3-linux-does-what-nintendont.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/07/31/blur-brutal-legend-ccra3-linux-does-what-nintendont.html</guid>
      </item>
    
      <item>
        <title>Updates and new apps for 2026</title>
        <description>PS1 Web Player</description>
        <content:encoded><![CDATA[<h2 id="ps1-web-player">PS1 Web Player</h2>

<p>While working on a new Playstation (1) game about my beagle, I realized that web emulation of PS1 games is a solved problem, and I shouldn’t be making folks use an emulater when a modern web browser will do.  I’ve added this to the Johnny Castaway PS1 project as well, so you can play that port and this game in your brower (tested on Android mobile as well)</p>

<ul>
  <li><a href="https://hunterdavis.com/johnny-castaway-ps1/play/online/">Johnny Castaway PS1 web emulator</a><br />
<a href="/content/images/2026/johnny-castaway-web-player-screenshot.png"><img src="/content/images/2026/johnny-castaway-web-player-screenshot-small.webp" alt="View the full-size Johnny Castaway PS1 web player showing Johnny on his island" width="320" height="214" /></a></li>
  <li><a href="https://hunterdavis.com/zoomies/play/">Zoomies PS1 web emulator</a></li>
</ul>

<h2 id="zoomies-ps1">Zoomies PS1</h2>

<p>I’ve been working on a fun-focused PS1 game, an infinite runner. It’s based on my real beagle, Munchie (Munchii in the game), it’s called Zoomies. It’s coming along nicely, with a fun rez-like mechanic where your health dictates the graphical fidelity of the game. It’s kind of the example game for a PS1 game engine I’ve been having fun building. It’s kind of just a silly game.</p>

<p><a href="/content/images/2026/zoomies-ps1-gameplay.webp"><img src="/content/images/2026/zoomies-ps1-gameplay-small.webp" alt="View the full-size Zoomies PS1 gameplay with Munchii running across rooftops" width="320" height="244" /></a></p>

<ul>
  <li><a href="https://hunterdavis.com/zoomies/">Zoomies PS1</a></li>
</ul>

<h2 id="dunking-sheep">Dunking Sheep</h2>

<p>Remember dunking bird?  I had made it multi-dunk able, but at the end of the day it no longer fit my workflow.  I love me some <a href="https://herdr.dev">herdr</a>. So I’ve re-created dunking bird to work directly via herdr’s socket interface and it’s a lot more direct.  Now you can auto-dunk on all of your agents across all of your herder panes.  Don’t trust your agent’s auto-time ability?  Want a centralised way to push all your various AI agents through to real gains without having to use yet another AI to orchestrate?  Dunk away!</p>

<p><a href="/content/images/2026/dunking-sheep-screenshot.png"><img src="/content/images/2026/dunking-sheep-screenshot-small.webp" alt="View the full-size Dunking Sheep interface selecting a Herdr agent pane" width="320" height="218" /></a></p>

<ul>
  <li><a href="https://github.com/huntergdavis/dunkingsheep">Dunking Sheep</a></li>
</ul>

<h2 id="tui-apps">TUI Apps</h2>

<ul>
  <li><a href="https://github.com/huntergdavis/media">Physical Media Editor</a><br />
<a href="/content/images/2026/physical-media-editor-screenshot.png"><img src="/content/images/2026/physical-media-editor-screenshot-small.webp" alt="View the full-size Physical Media Editor terminal interface showing board and card games" width="320" height="202" /></a></li>
  <li><a href="https://github.com/huntergdavis/openrouter-tui">ortop</a><br />
<a href="/content/images/2026/ortop-screenshot.png"><img src="/content/images/2026/ortop-screenshot-small.webp" alt="View the full-size ortop terminal dashboard showing OpenRouter usage charts" width="320" height="206" /></a></li>
  <li><a href="https://github.com/huntergdavis/dunkingbird">Dunking Bird</a><br />
<a href="/content/images/2026/dunking-bird-tui-screenshot.png"><img src="/content/images/2026/dunking-bird-tui-screenshot-small.webp" alt="View the full-size Dunking Bird terminal interface showing captured windows" width="320" height="145" /></a></li>
  <li><a href="https://github.com/huntergdavis/jellyterm">JellyTerm</a><br />
<a href="/content/images/2026/jellyterm-screenshot.png"><img src="/content/images/2026/jellyterm-screenshot-small.webp" alt="View the full-size JellyTerm terminal interface browsing a media library" width="320" height="145" /></a></li>
  <li><a href="https://github.com/huntergdavis/dunkingsheep">Dunking Sheep</a></li>
</ul>

<h2 id="more-projects">More Projects</h2>

<ul>
  <li><a href="https://github.com/huntergdavis/webtui">webtui</a><br />
<a href="/content/images/2026/webtui-screenshot.png"><img src="/content/images/2026/webtui-screenshot-small.webp" alt="View the full-size webtui browser interface booting a Debian terminal" width="320" height="214" /></a></li>
  <li><a href="https://github.com/huntergdavis/setupkde">setupkde</a><br />
<a href="/content/images/2026/setupkde-screenshot.png"><img src="/content/images/2026/setupkde-screenshot-small.webp" alt="View the full-size setupkde quick-start instructions for Kubuntu and Termux" width="320" height="242" /></a></li>
  <li><a href="https://github.com/huntergdavis/scamper-engine">Scamper</a><br />
<a href="/content/images/2026/scamper-engine-screenshot.png"><img src="/content/images/2026/scamper-engine-screenshot-small.webp" alt="View the full-size Scamper engine running Munchii through a platform arena" width="320" height="126" /></a></li>
</ul>
]]></content:encoded>
        <pubDate>Thu, 16 Jul 2026 10:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/07/16/updates-and-new-apps-for-2026.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/07/16/updates-and-new-apps-for-2026.html</guid>
      </item>
    
      <item>
        <title>Announcing Johnny Castaway PS1</title>
        <description>Introduction</description>
        <content:encoded><![CDATA[<h2 id="introduction">Introduction</h2>

<p>This has been a project well over a year in the making.  I’m proud to announce version 1.0 of Johnny Castaway PS1, a port of Johnny Castaway to the Playstation 1 console, with enhanced features and options like freeplay/sandbox mode. It’s the version of Johnny I always wished existed, and my most ambitious port to date.</p>

<h2 id="tldr">TLDR;</h2>
<p>If you’re less interested in the history and development of this project (which is an interesting case study in modern development, debugging, and optimization techniques), you can find the download and full website here: <a href="https://www.hunterdavis.com/johnny-castaway-ps1/">hunterdavis.com/johnny-castaway-ps1</a></p>

<p><img src="/content/images/2026/fishing1-ps1-cast.png" alt="Johnny Castaway PS1 fishing" width="640" /></p>

<h2 id="history">History</h2>

<p>I’ve ported the jc_reborn codebase to a few platforms over the years. I find porting enjoyable work, creative at times and necessitating compromise and cleverness. It was five years ago now when I released Johnny for the Dreamcast.  It was a bit of a slapdash port, back-porting jc_reborn to SDL1 for the DC. A year or two after it released, I receive an email from a game store owner who enjoyed it, but found that it crashed after about a half an hour (likely due to memory pressure from a large scene on the 8mb dreamcast)</p>

<p>I let them know I’d take a look.  Truth is, I hadn’t touched a DC toolchain in the years since, and re-creating it was a mountain to climb for someone with just an hour or two per week sometimes of free project time. Though I didn’t find the time, I thought about that email a lot.</p>

<p>Fast forward to 2025. It was (ironic, I know), AI coding agents that got me back into programming in my spare time.  The ability to parallelize my work with another developer (in this case an agentic one) brought some fun back into it for me.  I guess after all these years as an executive I’m a social butterfly. At any rate, I started to think about that email again.  I decided the best thing to do was write a version of jc_reborn with an extremely low memory footprint, then I could simply re-port to DC with SDL1 calls and solve the issue (if memory truly was the issue.)</p>

<p>That effort went very well and through research, hard work and an agentic partner we proved we could get the memory well below 8mb.  And that’s when that thing happens.  You know, that thing engineers do where they skip an easy problem for a harder one?  Yeah, I started to think about my large collection of gaming handhelds throughout the years. How many of them could run a dreamcast game? Only the newer ones, and not all of them.  How many though, could run a Playstation 1 game? Almost all of them, actually: from my first psp way back in 2004 to my steam deck and every phone for the past 15+ years.</p>

<p>Then it clicked for me, hard. The PS1 could do 640x480 interlaced. That’s Johnny’s native resolution. It had 600mb+ of space for assets. It’s been emulated, FPGA’d, and directly ran via chipset compatibility in the first four generations of playstation. Heck, maybe it could even run on the dreamcast via bleemcast?</p>

<p>So it began, in 2025.  First locally, as part of the jc_reborn project.  Then it grew in scope.  I wanted freeplay mode, background music, scene skipping, chapter selection, seed pinning, closed captions, memory card support. Everything anyone could possibly want in a version of Johnny for the playstation.  Something that actually could have come out at the time (maaaybe).</p>

<p><img src="/content/images/2026/activity9-ps1-boat.png" alt="Johnny Castaway PS1 waving at a boat" width="640" /></p>

<h2 id="the-journey-here">The Journey Here</h2>

<p>As with any major project, this one came in fits and starts. A straight port failed, with constant memory issues and minor bugs.  I spent months refining, before pivoting a few times into a methodology that actually worked.  A combination of host-pre-processing (essentially trading space for complexity) and relentless optimization.</p>

<p>I found it though.  That magical flow state that all software developers crave. After months and months of pain, custom tooling, hacking, AI after AI, I settled on a flow that brought me great joy and huge gains.</p>

<ul>
  <li><strong>Window 1</strong>: Dunking Bird, double dunking on OpenAI Codex and Anthropic Claude</li>
  <li><strong>Window 2</strong>: Claude, in an infinite editorial loop documenting progress on website/blog.</li>
  <li><strong>Window 3</strong>: Codex, in an infinite refinement loop regressing scenes against micro-optimizations in code and technique using headless docker containers.</li>
  <li><strong>Window 4</strong>: Fresh editor (used to use vim), for direct coding and code browsing</li>
  <li><strong>Window 5</strong>: Bottom monitor, keeping an eye on CPU, Memory, and Temperature</li>
  <li><strong>Window 6</strong>: Duckstation Emulator, running a visual regression for me</li>
</ul>

<p>All of this running on an old surface laptop (quite beat up, bought at thrift store), running KDE Neon, remote desktop’d in from my laptops/tablets/etc. A dream coding desktop always at my fingertips from anywhere.</p>

<p>I mean, think about it.  I’m coding up new features or fixing small bugs while two AI agents spend every minute of every day refining speed and documentation. This is all running on a machine costing less than 100$ and the pro tier of codex and claude.  Bonus: while I’m at home I connect a bluetooth controller to it and test freeplay mode and menus directly. While I’m far remote those get tested via headless session instrumentation.</p>

<p>It feels incredibly futuristic, and the proof is in the result.  In this single year I’ve been able to complete my most ambitious port yet, all while maintaining a full time job and significant family responsibilities.</p>

<h2 id="the-realness">The Realness</h2>

<p>There’s some AI slop/crap in the website.  It’s true.  A LOT. I would never allow such a thing on my personal blog (these are all my words, my feelings, I’m trying to convey something real). For a project site though, it’s really about presenting information and marketing. Very little emotion, much less need for the human touch. It’s also 100x better than the site I would have written for it.  You know my project sites: blog post, header image, links to download, credits.  That’s about it. This has feature articles, guides, how-tos, real documentation, technical articles that are actually kind of interesting and sometimes put an insightful spin on the technical challenges I faced.  These are not things I have the time to do nor the inclination most days.  Yet here we are with a really fun and informative project site. I’ll review what I can, write when I can, but when I can’t, I think this is a pretty interesting workflow for technical documentation.  An agent writes it as you struggle, not when you’re finished. It’s the struggle I find interesting, the challenge, the setbacks.  That’s the good part.</p>

<h2 id="the-desktop">The Desktop</h2>

<p>Here’s said workflow in action:</p>

<p><img src="/content/images/2026/dev-environment-2026-05-06-w1600.jpg" alt="Johnny Castaway PS1 development environment" width="600" /></p>

<p><a href="/content/images/2026/dev-environment-2026-05-06.png">Full Version</a></p>

<h2 id="getting-it-over-the-finish-line-in-an-emulator">Getting It Over The Finish Line In An Emulator</h2>

<p>This past couple of months, I’ve actually engaged in an even more aggressive triple-dunking setup. You see, performance and correctness are often at odds with each other. Throw in memory pressure, and you have a real iron triangle situation.  Any change to one affects the other two significantly. My strategy to get this over the finish line was essentially a phased re-prioritization of efforts.  In the beginning, my focus was purely on getting things running at all.  Towards the end of the project, I rotated that iron triangle in phases.  Correctness, then speed, then correctness again.  Once I had all scenes working perfectly, I rotated that triangle and focused on speed.  Once I had the speed optimized for all scenes, I rotated back focused on correctness again.  Bouncing back and forth, regressing stray pixels and timing issues. All the while, my third dunker worked on longevity testing.  The memory pressure side of the triangle.  Even a single byte leaked per scene will eventually build up to disaster.  It must run for days, if not weeks on end.  No 26-minute rule this time, no shortcuts.</p>

<p><img src="/content/images/2026/triple-dunking.png" alt="Johnny Castaway PS1 late stage triple dunking screenshot." width="600" /></p>

<h2 id="applicability-to-larger-projects-and-emulation">Applicability to Larger Projects and Emulation</h2>

<p>It strikes me that I’ve over-engineered this one.  Big surprise right, an engineer spending a year working on a passion project with no commercial value or intent reflects back on how it was over-engieneered.  I’ve sang that song too many times. Here’s the thing though that I think is both novel and valuable: the metaphor.  The analogue.  The next problem this can help solve. My mind goes to the field of emulation. You see, in one sense I hyper-optimized code on a scene-by-scene basis for one game. Well, that’s less a sense and more a literal interpretation.  Fuzz your eyes a little though.  63 scenes, each one with it’s own compression algorithms, loading schemes, techniques for memory pressure, saturizing the CPU, ensuring smoothness and correctness.  Keep fuzzing, but now imagine they are 63 individual games, all running within a shared emulator. Keep looking. Now start to expand the picture.  Github as the central repository for coordination.  Expand a bit more, now start to include more than a single developer and their AI agents.  What if it was hundreds, thousands of developers?  You see, the scaffolding I built for Johnny PS1 works equally as well for N64 games, or Dreamcast, or CDO. Except in those instances, there are thousands (millions) of folks willing to contribute.  And this scaffolding, it’s self-starting.  You need zero knowledge of development, docker, cross-compilers.  It can all be AI driven.  What if thousands of developers were dunking on improvements to emulation?  The possibility are endless, but the technique is straightforward. Ensure you have a harness that works not just in the visual domain, but on a systems and pressure level. There’s something incredible here, and I’m circling it, underlining it, and bolding it with yellow.</p>

<h2 id="further-parallelization-and-the-pursuit-of-perfection">Further Parallelization And The Pursuit of Perfection</h2>

<p>It was Memorial Day weekend 2026 when I finally got every scene into the 99%+ green speed category.  I had been struggling to get a few animation-heavy scenes over the finish line, and realized I was essentially running the same tests over and over again with small tweaks.  I had parallelized across agents, but I hadn’t saturated my local processing ability. Turns out a surface laptop 4 has no trouble running 20 simultaneous headless PS1 emulators for timing runs. So, naturally I wrote a script to generate worktrees in parallel and test small code changes tweaked via a predicate json config. You can see that script and the results in the worktree.</p>

<p>Will this port ever be at the point where I call it 100% perfect?  I don’t honestly know.  As I’m drafting this post we’re less than 0.3% slow for 20% of the scenes.  A third of a percent. Acceptable?  Probably.  Perfect? No, and I’ll continue to refine and update as I get the time.  Maybe someone else will too.  That’s my hope with this mountain of documentation and FAQs and regression suites etc etc.  It should be straightforward for anyone with interest to set up a development environment and try something out, and know exactly how that’s affected the runtime. Pretty cool, I think. In the meantime we’ll see how close to 100% I get this port for my planned V1 release mid-June. (Update from July, read on!)</p>

<h2 id="real-hardware-real-problems">Real Hardware, Real Problems</h2>

<p>The first week of July 2026 I was ready to do hardware testing. I picked up a chunky OG PS1 from the local gamestop for 60$ (somehow the cheapest option at the time.) I was determined to soak test Johnny on real hardware… and I did!  The first time a scene loaded it was glorious.  However, there were real problems to fix that didn’t manifest on an emulator.</p>

<ol>
  <li>The SPU (audio) init code wasn’t working correctly (timing) on the real hardware.</li>
  <li>Text rendering was missing (gpu pipeline and textures).</li>
  <li>Disk read failures regularly froze the game in a stall loop, a very common occurance with a burned CD and 30 year old hardware.</li>
  <li>The scene explorer preview image was garbled and would crash the game after a few previews.</li>
  <li>Some high-memory scenes would have paintover due to memory exhaustion (failure to pre-allocate background swap buffers.)</li>
  <li>Read failures during scene loads would cascade as issues in other systems.</li>
  <li>CD spin-up timing different on real hardware, leading to mis-timings.</li>
</ol>

<p>The truth is, I had designed this with the constraints of the hardware specifications of the PS1 in mind, but not the real world constraints of slowly degrading 30-year old hardware.  I’ve since hardened the read and retry code significantly.  Hard fails will eventually cause a scene swap or reload, but shouldn’t crash the system.  I’ve got decent performance out of a V1 PS1 and a burned CD launched from tonyhax/unirom boot, that’s the floor. Sometime this year I’ll try installing an ODE and see if that alleviates all of the read concerns and brings the performance in-line with emulators and FPGAs. Until then, we do what we can with the world we live in.</p>

<p>Not to say it wasn’t fun.  I did enjoy finding measures to alleviate these real world issues.  Those can be the most fun beacuse they tend to be the most interesting problems.  I started this holiday weekend with a stack of 50 blank cds (AZO ink) and they’re almost gone: sacrificed to the feiry pit of my burner.  It’s been years since I could say that and without reservation it brought me no small amount of joy. This release doesn’t mark the end of Johnny on the PS1, but the true beginning.  There’s a world of fixes and improvements that could be made, and I’ve no doubt in a few years LLMs (or better yet, a human with the time and drive) will be able to re-write the janky sprawling per-scene hackfest that is my codebase into something elegant and more performant. Until that time I’ll keep chipping away at bugs and small improvements. This project and indeed this past year in total have given me back my love of building things. My next game engine will for sure be on the PS1.</p>

<p>For now, please enjoy Johnny Castaway on the PS1.  I hope his story touches your heart as it did mine, I hope this gives you a reason to dust off your PS1 again, and I hope this project brings you a moment of quiet joy in this loud crazy world.</p>

<h2 id="try-it-out">Try It Out</h2>

<p><strong>GitHub Repository</strong>: You can find the complete source code, installation instructions, .bin/.cue files and documentation on GitHub: <a href="https://github.com/huntergdavis/johnny-castaway-ps1">huntergdavis/johnny-castaway-ps1</a></p>

<p><strong>Project Website</strong>: You can find all the above, and so much more on the official project website. 
<a href="https://www.hunterdavis.com/johnny-castaway-ps1/">hunterdavis.com/johnny-castaway-ps1</a></p>

<h2 id="have-some-fun">Have Some Fun</h2>

<p>This is designed to run forever (or as long as your machine can take it), perfect for emulators, virtual CD drives, FPGAs etc. My tests run for a week+ on emulators; no promises your PS1 won’t explode after being on that long.</p>

<p><img src="/content/images/2026/johnny1-ps1-the-end.png" alt="Johnny Castaway PS1 ending scene" width="640" /></p>

<h2 id="why-johnny-really">Why Johnny, Really?</h2>

<p>Johnny Castaway was the software I couldn’t afford on the computer I couldn’t afford in 1992.  By the time I had one that could run screensavers, they were far out of vogue. Still, when I was able to I ran Johnny on whatever I could.  The story really resonated with me.  Here’s someone trapped (perhaps a parallel to how you feel trapped in poverty), finding a way to have adventures and live life with very limited resources.  Making friends, finding love, and realizing what’s really important in life.  By the time he’s rescued and back to his programming job, he’s dreaming of the place he found happiness, back on that island. The struggle wasn’t the only thing he found on the island, he found himself and the way he wanted to live.  That’s always stuck with me, and the ending scene where he’s happily back on the island, that’s powerful. Jeff Tunnell created this little silly screensaver that’s also somehow profound. I’m thankful for his efforts, and every port is an homage to the original, and his vision.</p>

<p><img src="/content/images/2026/johnny6-ps1-date-dream.png" alt="Johnny Castaway PS1 dream date scene" width="640" /></p>

<hr />
]]></content:encoded>
        <pubDate>Sun, 05 Jul 2026 10:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/07/05/announcing-johnny-castaway-ps1.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/07/05/announcing-johnny-castaway-ps1.html</guid>
      </item>
    
      <item>
        <title>The Promise of Distributed Optimization in Software Development</title>
        <description>Introduction</description>
        <content:encoded><![CDATA[<h2 id="introduction">Introduction</h2>

<p>It’s no secret I’ve been working on <a href="https://www.hunterdavis.com/johnny-castaway-ps1/">something cool and unexpected</a>, with a big release announcement coming very soon. However during the course of development I stumbled upon a pattern which I think could provide huge dividends for the emulation community, and is very straightforward to implement.</p>

<h2 id="tldr">TLDR;</h2>
<p>A full development environment, agentic performance improvement harness, and control plane is just a click away for the average user. How far can we push our open-source projects now that truly anyone can contribute? What if every game, every function, every aspect of our software was being continually optimized by each user, in a distributed fashion?</p>

<h2 id="building-the-flow">Building The Flow</h2>

<p>Each part of this flow came from an actual need during development of the Playstation 1 port of Johnny Castaway.  That means you can go over and try this today, and I’ve had it successfully working for many months.</p>

<ul>
  <li><strong>An Execution Harness</strong>: You need some way to continually execute experiments.  For me that was a claude or codex loop, Dunking Bird (obviously before the /goal feature dropped), running headless on a server in my closet. This could be a human, but the point of this article is that everyone can now do this, not just those of us professionally trained to do so.</li>
  <li><strong>A Reasoning Engine</strong>: You need a way to reason about the code, postulate improvements, and execute those improvements within code.  Again for me this was claude, codex, ds4, etc.</li>
  <li><strong>A Centralized Log</strong>: Github or any distributed version control system. You need a log of ideas, a log of successful experiments, and a log of failed experiments.</li>
  <li><strong>A Regression Suite</strong>: You need some way to regress performance or ground truth against your experiment. Deep performance logging. For Johnny, that was deep performance timing measurement emitted from a known-good port to x86 systems.</li>
  <li><strong>A Finite, Decomposable Problem Set</strong>: Finally, you need a problem set or software architecture that’s decomposable.  Otherwise you’ll waste effort with everyone re-implementing the same area as folks work to improve the system in parallel. Johnny had 127 unique scenes, each with its own memory profile and asset usage. I suppose it’s not a strict requirement but I’m saying this technique is for ‘problem sets’</li>
</ul>

<p>Simple parts coming together to deliver on one of the central promises of open source: that everyone can improve software together. For the port I’ve been working on, you can have a new node up and running in a few minutes with any off-the-shelf agentic model. Perf testing, improvement looping, dockerized headless runs, dockerized cross-compiler and iso builder. Anyone can run, and anyone can contribute.</p>

<p>An agent in an infinite refinement loop regressing against micro-optimizations in code and technique using headless docker containers.</p>

<p><img src="/content/images/2026/perflog.png" alt="The Performance Improvement Log" width="600" /></p>

<p>All of this runs on an old surface laptop (quite beat up, bought at thrift store), running KDE Neon, remote desktop’d in from my laptops/tablets/etc. A single node in an ever-growing distributed system, each node doing it’s part to improve the system.</p>

<p>It feels incredibly futuristic, and stupid simple at the same time.  In the past month using this technique I’ve taken speed from 70% to 99.8% average in an incredibly difficult port with little to no headroom. Thousands of tiny improvements stacking on each other for real gains.</p>

<h2 id="the-realness">The Realness</h2>

<p>There’s some AI slop/crap in the code still.  Certainly there’s a lot of cleanup I can do, it’s not a particularly clean codebase.  It’s not great for pedagogy, or human maintainability. Those are open, hard problems I’m working to solve. What it is, is working well.  It’s the impossible port I didn’t have enough time to get working at full speed on my own. As I take a step back and start to prepare for a version 1 release, I realize it’s a prototype for a paradigm that threads through my entire career.  Distributed computing. Open source. Optimization. Bringing new ability to the non-technical contributor.</p>

<p>I think back to high school, 30+ years ago. How I looked in awe upon the game developers who released their code, the very early emulator developers who seemed to do impossible things.  I wanted to help, to be a part of it.  Those are some of the feelings that drove me into a career in computer science, into who I am today.  Would I still have taken that path today if there was an ‘easy button’ to contribution?  Maybe. Probably. But that’s not the problem I’m out here trying to solve, that’s not the world I’m out here dreaming in public.</p>

<h2 id="bringing-it-all-together-for-emulation">Bringing it all together for Emulation</h2>

<p>There are trends in our industry that seem to be converging, and that ever-present nebula will soon fold into a black hole in my mind. It’s hard to divorce my thinking from the reality of the world we’re in today.  Memory shortages are driving the price of memory sky-high and the specs on the average computing device have started to fall. The MacBook Neo launched to great acclaim with 8gb of RAM. At the same time frontier models have continued to climb in capacity, currently meeting or exceeding the performance of the average junior to mid-level engineer.</p>

<p>All the while, I think about performance. I’m very much aligned with John Carmack: the world could run our current software on hardware that’s an order of magnitude less powerful if we actually cared about performance optimization.</p>

<p>And there’s the wind-up. Now imagine you’re, say, an emulator developer.  You don’t want to deal with a billion incoming pull requests, AI slop, noise, constant stress. So you carve off something. Maybe it’s a per-game setting of some sort.  Maybe you’re trying to do something a little ridiculous. So you say (much as I did), that there’s a huge amount of opportunity in custom compression/decompression algorithms, and in saving decoding time. So you open it up, each game has it’s own configuration file detailing compression and decompression schemes and their implementation, maybe pre-compiled and loaded statically at runtime.  Doesn’t matter.</p>

<p>Now every user who can afford 10$/mo in Deepseek usage can contribute. What percentage of your audience would contribute knowing that it’s just a button press away? I as one node was able to implement thousands of micro-improvements over 120 scene variants in a couple of months.  With 100 users doing the same it would have taken days. What could 1000 do?  A million? How far could we push even a single optimization path with a billion experiments?</p>

<p>What can we do when the barrier to entry is effectively zero? (admitting that there have been times that 10$/mo felt like an infinite amount, and recognizing my current privilege to be able to say so.)  Knowing projects can be structured in this way today, I’m emboldened.</p>

<h2 id="the-desktop">The Desktop</h2>

<p>Here’s a screenshot of the workflow in action:</p>

<p><img src="/content/images/2026/dev-environment-2026-05-06-w1600.jpg" alt="Johnny Castaway PS1 development environment" width="600" /></p>

<p><a href="/content/images/2026/dev-environment-2026-05-06.png">Full Version</a></p>

<h2 id="a-simple-distributed-optimization-template-for-my-next-project">A Simple Distributed Optimization Template For My Next Project</h2>

<ol>
  <li>However it’s built, that must happen in docker from day 1.</li>
  <li>However it runs, it must be able to run headlessly, preferably in docker.</li>
  <li>Wherever the source is stored, it must be in a distributed fashion, and contain 4 logs: performance targets vs current, improvements, successful experiments, failed experiments.</li>
  <li>Whatever it does, there should be clear performance measurement and targets, stored within the source.</li>
  <li>Whatever those targets are, they should be distinct and decomposable.</li>
</ol>

<h2 id="try-it-out">Try It Out</h2>

<p><strong>GitHub Repository</strong>: You can find the complete source code, installation instructions, and documentation on GitHub: <a href="https://github.com/huntergdavis/johnny-castaway-ps1">huntergdavis/johnny-castaway-ps1</a></p>

<p><strong>Project Website</strong>: You can find all the above, and some interesting write-ups (and some AI crap) on the official project website.
<a href="https://www.hunterdavis.com/johnny-castaway-ps1/">hunterdavis.com/johnny-castaway-ps1</a></p>

<h2 id="finally">Finally</h2>

<p>I’d love to hear if anyone else has tried this (very similar concept to autoresearch.)</p>

<hr />
]]></content:encoded>
        <pubDate>Wed, 27 May 2026 10:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/05/27/the-promse-of-distributed-optimization.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/05/27/the-promse-of-distributed-optimization.html</guid>
      </item>
    
      <item>
        <title>Dumber and More Dangerous than Ralph Wiggum: The Dunking Bird Method</title>
        <description>Introduction</description>
        <content:encoded><![CDATA[<h2 id="introduction">Introduction</h2>

<p>Do you remember that episode of the Simpsons where he uses a little dunking bird toy to defraud his employer? Yeah. Like that but with prompts.  Here’s the thing. Some of my coding agents, no matter how “dangerously” I like to live, still stop every 10 minutes, needing my input to continue. I know Ralph Wiggum method doesn’t have this problem, but that supposes you have planned out your project to some degree. Also I’m lazy, and it is easier ask AI to write a tool to do a ridiculous thing than to figure out another ridiculous thing.</p>

<p>If I’m really living the agentic dream, why don’t my agents work while I’m dreaming?</p>

<h2 id="what-is-dunking-bird">What is Dunking Bird?</h2>

<p>Dunking bird is my metaphor for programs which sit outside the interface of agents and act on them externally, repeatedly sending a set prompt until a completion state is reached. It’s also a sample implementation of the concept I coded up today.</p>

<h2 id="what-does-it-do">What does it do?</h2>

<ul>
  <li><strong>Feature 1</strong>: Click-to-capture a window (tested on KDE/wayland)</li>
  <li><strong>Feature 2</strong>: Prompt at a settable interval (1. ; continue; yes that’s a great idea!, etc )</li>
  <li><strong>Feature 3</strong>: test button for confidence</li>
  <li><strong>Feature 4</strong>: Insecure!</li>
</ul>

<h2 id="screenshot">Screenshot</h2>

<p>Here’s Dunking Bird in action:</p>

<p><img alt="" src="/content/images/2026/dunking-bird-screenshot.png" width="600" /></p>

<h2 id="the-story-behind-dunking-bird">The Story Behind Dunking Bird</h2>

<p>I’ve been working on a PS1 version of Johnny Castaway for about a year now.  I have agents doing static analysis and resource conversion, and it’s a pain to have to type “continue” every 10 minutes or so. To that end, I vibe-coded up a tool that’d work for my development environments.</p>

<h2 id="try-it-out">Try It Out</h2>

<p><strong>GitHub Repository</strong>: You can find the complete source code, installation instructions, and documentation on GitHub: <a href="https://github.com/huntergdavis/dunkingbird">huntergdavis/dunkingbird</a></p>

<h2 id="ai-slopportunity">AI-Slopportunity</h2>

<p>Sloppity slop slop, dropitty drop drop look at AI go</p>

<p><img alt="" src="/content/images/2026/dunking-bird-ai-art.png" width="400" /></p>

<hr />
]]></content:encoded>
        <pubDate>Sat, 21 Mar 2026 10:00:00 +0000</pubDate>
        <link>https://hunterdavis.com/2026/03/21/announcing-dunking-bird.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2026/03/21/announcing-dunking-bird.html</guid>
      </item>
    
      <item>
        <title>A big ole stack of one-off projects in 2025</title>
        <description>I actually wrote quite a bit of code in 2025! While my professional life was centered around AI enablement for my teams and getting the big launch out this year to customers, I did a fair amount of personal projects.</description>
        <content:encoded><![CDATA[<p>I actually wrote quite a bit of code in 2025!  While my professional life was centered around AI enablement for my teams and getting the big launch out this year to customers, I did a fair amount of personal projects.</p>

<p>So this is just a quick hits list for 2025 of some fun personal projects I worked on, all of which have source available, are local-first and do not require a back-end. Mostly javascript, always client-side. Many were written with signifant AI enablement, I’m finding that my AI workflow anymore is comparible to those few years I was leading R&amp;D as an IC developer, only instead of interns and graduate students working with me to refine prototypes, it’s AI. Is it good for our profession? I honestly don’t know, but I doubt it. Is it incredibly fun and rewarding? Yes, 1000% yes.</p>

<h2 id="picocalc">PicoCalc</h2>
<p><img alt="" src="/content/images/2025/picocalc.png" width="400" /></p>

<p>Back in April, I was messing with the default basic interface on my picocalc (pi pico h1), and wanted a better file browsing shell interface.  You can find that here:
<a href="https://github.com/huntergdavis/picocalc">Github Repo</a></p>

<h2 id="inboxzero">InboxZero</h2>
<p><img alt="" src="/content/images/2025/inboxzero.png" width="400" /></p>

<p>Around the same time, I took about 30 seconds to code up a button that throws confetti and plays trumpets, a more visceral inbox zero. Hardly worth mentioning, but again fun :)</p>

<p><a href="https://github.com/huntergdavis/inboxzero">Github Repo</a>
<a href="/inboxzero">Live Link</a></p>

<h2 id="streak">Streak</h2>
<p><img alt="" src="/content/images/2025/streak.png" width="400" /></p>

<p>I’ve had significant health issues this year, and this summer I was struggling with going to the gym every day. I wrote this little app ‘streak’, it bascially takes (or generates) your query parameter and sets a cookie for whenever you press the button, keeping a running log of when you did that thing. Celebratory confetti if you did it less than 1 day prior.</p>

<p><a href="https://github.com/huntergdavis/streak">Github Repo</a>
<a href="/streak">Live Link</a></p>

<h2 id="visualizer">Visualizer</h2>
<p><img alt="" src="/content/images/2025/visualizer.png" width="400" /></p>

<p>I loved winamp 25 years ago.  LOVED IT.  All my music (mp3s), customization, but most of all visualizers!!  I put a record on this summer, and realized that I didn’t have an easy way to load up winamp quickly on whatever system I was on.  So this is just a microphone-input visualizer that supports all of winamps visualization plugins via milkdrop.  Again, client-side only, all javascript.  I find myself using this one a lot actually.</p>

<p><a href="https://github.com/huntergdavis/visualizer">Github Repo</a>
<a href="/visualizer">Live Link</a></p>

<h2 id="solitaire">Solitaire</h2>
<p><img alt="" src="/content/images/2025/solitaire.png" width="400" /></p>

<p>Remember back in Covid, I designed that solitaire variant? I wrote a simulator for it to help play-test and balance, but folks wanted to actually play it.  So, that’s what this is, a webapp to play my weird solitare variant.</p>

<p><a href="https://github.com/huntergdavis/solitaire">Github Repo</a>
<a href="/solitaire">Live Link</a></p>

<h2 id="people-grid">People Grid</h2>
<p><img alt="" src="/content/images/2025/peoplegrid.png" width="400" /></p>

<p>A good amount of my job in engineering leadership is around team formation, run rate management, and organizational design.  When I want to scratch something up real quick, figma is a pretty good tool, but I don’t like to pay for simple tools with too many features, or large corporations. People grid is just a simple org design tool that I use to sketch up team formations and simple team diagrams.  A little rough around the edges, but I end up using this one all the time.</p>

<p><a href="https://github.com/huntergdavis/peoplegrid">Github Repo</a>
<a href="/peoplegrid">Live Link</a></p>

<h2 id="poke-bar-game">Poke (Bar Game)</h2>
<p><img alt="" src="/content/images/2025/poke.png" width="400" /></p>

<p>A good buddy of mine had an idea for a bar game.  It tests your reflexes, then always tells you that you’re too drunk to drive.  Cheeky right? We did it up fun with some graphics and delighters.</p>

<p><a href="https://github.com/huntergdavis/poke">Github Repo</a>
<a href="/poke">Live Link</a></p>

<h2 id="web-flight-flight-sim-concept">Web Flight (flight sim concept)</h2>
<p><img alt="" src="/content/images/2025/webflight.png" width="400" /></p>

<p>I always loved flying games, especially arcade ones.  Superflight is one of my favorite casual games ever made.  I started on a basic web flying game, but later pivoted the idea into asteroid miner. Here’s the original webflight prototype.</p>

<p><a href="https://github.com/huntergdavis/webflight">Github Repo</a>
<a href="/webflight">Live Link</a></p>

<h2 id="psyrunner-mobile-runner-game">PsyRunner (mobile runner game)</h2>
<p><img alt="" src="/content/images/2025/psyrunner.png" width="400" /></p>

<p>I miss flash games, and only feel that lately javascript development is starting to feel like what simple flash games development felt like 20 years prior. There was one, it was kind of based on the forest of endor chase, where you would avoid triangles in a straight line.  This is a modern update to that format, with joystick, mouse, and mobile touch controls so you can play on your phone. Actually pretty fun!</p>

<p><a href="https://github.com/huntergdavis/psyrunner">Github Repo</a>
<a href="/psyrunner">Live Link</a></p>

<h2 id="asteroid-miner-3d-asteroid-mining-arcade-sim">Asteroid Miner (3d Asteroid Mining Arcade Sim)</h2>
<p><img alt="" src="/content/images/2025/asteroidminer.png" width="400" /></p>

<p>I have put so, so many hours into No Man’s Sky. That’s a team that continues to impress, and I keep coming back.  Ironically, most playthroughs I start to build-up, then get distracted and go asteroid mining.  Flying through space, busting asteroids for minerals, upgrading my ship.  It’s just a fun casual game loop.  So, I’ve started to work on a game that is just that, 3d space mining.  Still some rough edges, but by far the most impressive/complex project I worked on personally in 2025. Full joystick or keyboard/mouse controls, graphics are pretty decent, and I make regular updates on Fridays.</p>

<p><a href="https://github.com/huntergdavis/asteroidminer">Github Repo</a>
<a href="/asteroidminer">Live Link</a></p>

<h2 id="2d-js-game-boilerplate-boilerplate-for-2d-javascript-webgames">2d JS Game Boilerplate (boilerplate for 2d javascript webgames)</h2>
<p><img alt="" src="/content/images/2025/2djsgameboilerplate.png" width="400" /></p>

<p>Ever have a game that really sparks your creativity?  Donkey Kong Bananaza was that game for me in 2025. I have so many 2d game concepts I’m going to try out, just based on fun little mechanics they threw in.  To start, I’ve created a simple boilerplate for 2d games client-side javascript (you know… stuff like bounding box, mini-map, joystick support and movement controls.)</p>

<p><a href="https://github.com/huntergdavis/2djsgameboilerplate">Github Repo</a>
<a href="/2djsgameboilerplate">Live Link</a></p>

<p>So, last year I committed to releasing more things that spark joy, and more little projects. I remembered to make space to work on them, but only now am I actually releasing some stuff into the wild. Hey, progress is progress :D</p>

]]></content:encoded>
        <pubDate>Sun, 05 Oct 2025 01:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2025/10/05/a-bunch-of-one-off-projects-for-2025.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2025/10/05/a-bunch-of-one-off-projects-for-2025.html</guid>
      </item>
    
      <item>
        <title>Announcing TPS</title>
        <description>Announcing TPS (Team Planning Simulator) (see, it generates TPS Reports ha!)</description>
        <content:encoded><![CDATA[<p>Announcing TPS (Team Planning Simulator) (see, it generates TPS Reports ha!)</p>

<p>Source and download here -&gt; <a href="https://github.com/huntergdavis/teamplanningsimulator/">TPS</a></p>

<p>And hey, as it’s client-side, how about a <a href="/teamplanningsimulator/">Live Version</a></p>

<p><img alt="" src="/content/images/2024/tps.png" width="400" /></p>

<p>Team Planning Simulator Reports (TPS Reports) is a quick report generator to help you plan hiring, throughput, and costs for the year, broken down by quarter and 2-week sprint. And yeah, the name is a riff on the ole T.P.S. report.</p>

<p>Current features: 
So few, so laughably few.  But hey, enough to honestly help in some situations.</p>

<ol>
  <li>Save and load to JSON for fast iteration!</li>
  <li>Add roles and event types, then apply them to your calendar</li>
  <li>See throughput metrics, efficiency, costing, etc.</li>
</ol>

<p>It falls squarely into the bucket of “can’t you just use a spreadsheet?”  Yes. Of course you can, I’ve done so for a few decades now and I’m sure folks will do so for another few hundred. The humble spreadsheet will outlive us all. All that said, this is a helpful way for me to visualize, quickly iterate with small features, and quickly generate scenarios without relying on a cloud-hosted python backend for a simulator or stacks and stacks of resource-crazy google sheets.</p>

<p>Backstory: 
I’m committing to releasing more things that spark joy, and more little projects.  Sometimes you just write a quick simulator on a Sunday night to help you clear your head and get your thoughts out, and sometimes that’s really helpful to other folks.</p>

]]></content:encoded>
        <pubDate>Sun, 29 Dec 2024 01:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/12/29/announcing-tps.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/12/29/announcing-tps.html</guid>
      </item>
    
      <item>
        <title>Announcing Labrync</title>
        <description>Announcing Labrync (like labrynth + ncurses)</description>
        <content:encoded><![CDATA[<p>Announcing Labrync (like labrynth + ncurses)</p>

<p>Source and download here -&gt; <a href="https://github.com/huntergdavis/labrync/">Labrync</a></p>

<p><img alt="" src="/content/images/2024/earlygame.png" width="400" /></p>

<p>Labrync is a passive or active little console toy. I wanted to play with a few maze exploration algorithms (you can see a bunch in the history).  The color palette is going to change as right now it’s a ‘hotdog stand’ level thing. I thought the concept from onelonecoder was really fun and inspiring and wanted to extend it in some fun ways.  Also ported to python cause why not I guess?</p>

<p>Like I said before, little things that bring me joy.  Things I might dip in and add a feature to once a year, for 30 years. That’s what my internet is all about.  Here’s another one.</p>

<p><a href="https://github.com/user-attachments/assets/89f0045f-518f-4b25-9698-34922285b47d">Early Demo Movie</a></p>

<p>Based on and extended from the command line C fps from 
onelonecoder</p>

<p><a href="https://github.com/OneLoneCoder/CommandLineFPS">One lone coder Github</a></p>

<ol>
  <li>Maps are randomly generated</li>
  <li>Maps can be auto-solved (screensaver mode, if you will) by passing -a to labrync.py</li>
  <li>Colors are… supported</li>
  <li>Player entering ‘X’ exit will regenerate the map and start fresh in top left corner</li>
  <li>Breadcrumbs (or .. reverse breadcrumbs?  Crumpled grass?) To show where you’ve been.</li>
  <li>Fog of War (or not) for additional challenge</li>
</ol>

]]></content:encoded>
        <pubDate>Sat, 16 Nov 2024 01:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/11/16/announcing-labrync.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/11/16/announcing-labrync.html</guid>
      </item>
    
      <item>
        <title>Announcing Tui000</title>
        <description>Announcing Tui000 (sounds like ‘three thousand’)</description>
        <content:encoded><![CDATA[<p>Announcing Tui000 (sounds like ‘three thousand’)</p>

<p>Source and download here -&gt; <a href="https://github.com/huntergdavis/tui000/tree/main">Tui000</a></p>

<p>Tui000 is a passive game / screensaver for your terminal.  Watch virtual avatars make life decisions, visit the graveyard to see how each life measured out. Passive, small. Now is the time for small joys, for silver linings. I feel 2025 will need as much joy as possible, and I’ve been inspired to put more little moments of joy out there for folks like I used to.</p>

<p>I had an idea one day just relaxing and thinking about Johnny Castaway. 
I had already ported Johnny to as many platforms as I was interested in.
What I really missed was that feeling of fun and writing software.</p>

<p>Some of my best work was made famous not by my original intention.
Rather, it was the creativity of others that brought purpose to “meaningless” software. 
After all these years, I keep coming back to the terminal.</p>

<p>So, in an attempt to recreate that feeling, in the terminal, here’s the start and the pitch.
You only have 3000 weekends, on average, in your adult life. The life-map assumes that each major choice you make in your life will become the focus, on average, of 12 weekends. Some a lot more, some a lot less, but on average let’s take a wild swing at 12 or so.</p>

<p>So you get 240 decisions, some many times over, with how to spend your time.  Sounds like a lot, goes by quick.  Just like life. Probably not of interest to most folks, but of great interest to me.  Useless, silly, simplistic, exactly what I had envisioned.  All future improvements are fun additions to the base concept, and I’ve got a ton in mind.</p>

<p>Design:</p>

<p>Constraints: Must target / be usable in 80x24 terminal window.</p>

<p>Launching the app spawns a character.
The character makes choices, indicated by colors which map to life categories. 
These choices add up to weave the tapestry of each life, the colorful headstone.
When the character dies, each is written out to a json file in ./graveyard/</p>

<p>Running the App:
Run with python after installing requirements.txt</p>

<p>Pass the -debug flag for more logging and 100x speed</p>

<p>Building with Docker:
sudo docker build -t tui000 .</p>

<p>Running with Docker:
docker run -it –name tui000-container tui000</p>

<p>You can find the repository below, including original design docs and screenshots.</p>

<p><a href="https://github.com/huntergdavis/tui000/tree/main">Tui000</a></p>

<p>Main Graveyard Screen -&gt; 
<img alt="" src="https://github.com/huntergdavis/tui000/raw/main/screenshots/graveyard_like_concept.png" width="400" />
<img alt="" src="https://github.com/huntergdavis/tui000/raw/main/concept_art/death_screen.jpg" width="400" /></p>
]]></content:encoded>
        <pubDate>Thu, 07 Nov 2024 01:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/11/07/announcing-tui000.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/11/07/announcing-tui000.html</guid>
      </item>
    
      <item>
        <title>Client-Side Site Search for HunterDavis.com</title>
        <description>I stumbled upon an Excellent Blog Post, from Stephan Miller detailing how to set up a local client-side search functionality for static sites.</description>
        <content:encoded><![CDATA[<p>I stumbled upon an <a href="https://www.stephanmiller.com/static-site-search/">Excellent Blog Post</a>,  from Stephan Miller detailing how to set up a local client-side search functionality for static sites.</p>

<p>So that’s exactly what I implemented today. You’ll see a search box on the right of the homepage, and you can sort the results by relevance or date. Best of all, this relies on no external search engine and respects your privacy. This works by generating a json blob with all my post data, then running a local search with lunr.js.  Much, MUCH easier to find content now, especially those old Zipit and game review articles.</p>

<p><img alt="" src="/content/images/2024/sitesearch.png" width="400" /></p>
]]></content:encoded>
        <pubDate>Sun, 25 Aug 2024 12:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/08/25/search.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/08/25/search.html</guid>
      </item>
    
      <item>
        <title>Physical Media and Markdown Table Renderer</title>
        <description>I’m a big fan of Obsidian, paying for premium and using it to store notes and lists.</description>
        <content:encoded><![CDATA[<p>I’m a big fan of <a href="https://www.obsidian.md/">Obsidian</a>, paying for premium and using it to store notes and lists.</p>

<p>I was enjoying myself rather thoroughly this weekend while browsing a thrift store in sunny Montana, a couple states away from rainy Seattle. There was a great deal of most excellent used DVDs, Blu-Rays, VHS tapes and other media formats I still find myself using day to day.</p>

<p>While I have obsidian sync for my local devices, I pondered how to keep my family apprised of what media we have and what we don’t, especially when that number is quickly approaching the thousands from the hundreds?</p>

<p>There’s a publish option, but that’s an additional 10$ per month and I already have a website and GH pages. I realized what I really needed was a way to render markdown tables as HTML client-side, then I could simply point a static page to those markdown files and I’ve got a nice little interface on the web.</p>

<p>So that’s exactly what I wrote up today. <a href="/media/">Live Link.</a> <a href="https://github.com/huntergdavis/media">Github Link.</a></p>

<p><img alt="" src="/content/images/2024/markdown.png" width="400" /></p>
]]></content:encoded>
        <pubDate>Sun, 25 Aug 2024 12:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/08/25/media.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/08/25/media.html</guid>
      </item>
    
      <item>
        <title>2024 Privacy and AI Update</title>
        <description>April Cool’s Day inspired me to do something different and share something unexpected.</description>
        <content:encoded><![CDATA[<p><a href="https://www.aprilcools.club/">April Cool’s Day</a> inspired me to do something different and share something unexpected.</p>

<p>We live in a very different world from the one that existed when this blog was created. It’s taken many different forms over the years, but throughout it all my desire to share and help others remains unchanged.</p>

<p>Where do I draw the line as to how far that help extends? When I think of my articles being helpful to a new leader or aspiring hacker, that fills me with joy. When I think of large companies using my likeness and writing style to make billions while trashing our environment, I pause. When I think of the bad actors, the scammers, the phishers, the AI-enabled stalkers (all of whom I’ve experienced over these past couple of years), I fear. I fear that they’ll continue to use the good I’ve tried to share with the world to hurt myself and others.</p>

<p>Fear can be a powerful motivator for change, but so too can the desire to help others. I want to set a better example here, when I write I want to write about leadership and coaching with a more inclusive voice and patterning better behaviors. Much as I have done in my professional life I’m going to start leveling up my game here on the blog. Today that means a couple of changes to the site, the removal of some content, and a promise to take a silver lining from this refresh.</p>

<p>Change number 1: I’m explicitly licensing all sources on this blog going forward with a NON-AI MIT license.</p>

<p>Change number 2: The removal of overly personal blog posts and content. While photos of my family and pets from 20 years ago still bring a tear to my eye, they pose a risk to their safety in this new world.</p>

<p>Change number 3: The de-identification of personal imagery and personally identifying information from all articles and site pages. Going to have some fun with this one, adjusting some prior images algorithmically. For now, I’ll start with <a href="/content/images/2024/imagechange.py">a quick thresholding and random colorization.</a> Moving forward: I’ll find a new avatar and remove EXIF data from any photos.</p>

<p>Here’s what this quick first pass looks like:
<img alt="" src="/content/images/2013/08/20130830_144712.jpg" width="400" /></p>

<p>Silver Lining: One great lesson I’ve learned in my career, imparted to me by a leader I trust: not everyone you lead will see themselves in you. The sames goes for readers of this blog! My hope is that more folks may connect with the message I’m trying to send going forward.</p>
]]></content:encoded>
        <pubDate>Mon, 01 Apr 2024 12:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2024/04/01/privacy-and-ai-update.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2024/04/01/privacy-and-ai-update.html</guid>
      </item>
    
      <item>
        <title>My Custom Guitar from MaCo Guitars Is In!</title>
        <description>This has been many years coming, and I’m so excited to finally be able to share this with y’all. Years ago, I met a brilliant engineer and guitar builder, who later started MaCo guitars. Many years before thats, my co-founder Mark and I had built a cigar-box guitar together based on the Make magazine cover article. I never forgot how much fun it was, or how much skill goes into even the most basic of guitar builds. I knew, one day I would commission a professional shop to build my dream guitar.</description>
        <content:encoded><![CDATA[<p>This has been many years coming, and I’m so excited to finally be able to share this with y’all.  Years ago, I met a brilliant engineer and guitar builder, who later started <a href="http://www.macoguitars.com/">MaCo guitars.</a> Many years before thats, my co-founder Mark and I had built a <a href="/about.html#guitar_">cigar-box guitar</a> together based on the Make magazine cover article. I never forgot how much fun it was, or how much skill goes into even the most basic of guitar builds.  I knew, one day I would commission a professional shop to build my dream guitar.</p>

<p>Well dear readers, that day has come. Below, learn all about my new guitar, and how it came together.  Please note, this is NOT a paid advertisement. I have received no compensation or discounts for writing this article, I’m just so frikkin amazed and in love with my guitar, I want everyone to know!</p>

<p>I reached out to my friend Mat at MaCo, and he walked me through the whole process. It started with a shared design session, with Mat and I going through the customizations, shape, materials, and wood types.  This was easy, and a bit like putting a computer together online, only in this case everything was custom, and changable.</p>

<p>Over the next year, I received a regular stream of updates from MaCo.  The wood selection.  The resonator build.  The composite materials and cuts.  They became the highlight of my year, and certainly some of the best and most engaging news I had in 2021.</p>

<p>Then, at the end of 2021 it was ready!  I could hardly contain my excitement!  I flew out to pick up this one of a kind guitar, and I was not ready for how emotional it would be opening the case for the first time!</p>

<p><img alt="" src="/content/images/2022/opencase.jpeg" width="400" /></p>

<p>The beauty and complete uniqueness of my guitar still amaze me all these months later. I was nearing tears as I was presented with the build certificate.</p>

<p><img alt="" src="/content/images/2022/certificate.jpeg" width="400" /></p>

<p>I had a particular set of (perhaps odd, definitely non-standard) requirements for this build.  I wanted a shortened neck, with a semi-hollow body and a player-reflecting resonator.</p>

<p><img alt="" src="/content/images/2022/guitar_body.jpg" width="400" /></p>

<p>Look at those tuning pegs!  Oooh they are pretty.</p>

<p><img alt="" src="/content/images/2022/head.jpg" width="400" /></p>

<p>My hands were literally shaking as I picked up the guitar for the first time.  Lightweight and perfectly balanced, with that custom cut MaCo is known for.</p>

<p><img alt="" src="/content/images/2022/holdingguitar.jpeg" width="400" /></p>

<p>I was giddy as I looked into the resonator and saw the build certificate and born-on date!!  That’s my name in there!</p>

<p><img alt="" src="/content/images/2022/inthebody.jpg" width="400" /></p>

<p>MaCo even took some of the leftover premium woods and created a custom set of coasters for me.  So thoughtful!</p>

<p>One last shot of my guitar.</p>

<p><img alt="" src="/content/images/2022/righthand.png" width="400" /></p>

<p>Thanks for reading all about my guitar!!  I highly recommend anyone interested in top tier professional guitar-builders to check out MaCo guitars!  It’s been a life-changing experience for me, the creation of an heirloom I’ll treasure for life!</p>
]]></content:encoded>
        <pubDate>Sun, 03 Apr 2022 12:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2022/04/03/custom-guitar-from-maco.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2022/04/03/custom-guitar-from-maco.html</guid>
      </item>
    
      <item>
        <title>Coaching Leaders During Covid</title>
        <description>It’s 2022. Strange as it is to write this, that means we’ve been living in a Covid-19 world for two years. For many of us that has meant a shift into remote work and remote leadership. Looking back on these past two years of fully remote leadership, I’ve observed that the stress of Covid-19 has only exacerbated the struggles that many new leaders go through. With that in mind, here are four struggles new leaders commonly face, and coaching lines I tend to employ for each.</description>
        <content:encoded><![CDATA[<p>It’s 2022. Strange as it is to write this, that means we’ve been living in a Covid-19 world for two years. For many of us that has meant a shift into remote work and remote leadership. Looking back on these past two years of fully remote leadership, I’ve observed that the stress of Covid-19 has only exacerbated the struggles that many new leaders go through. With that in mind, here are four struggles new leaders commonly face, and coaching lines I tend to employ for each.</p>

<p>Please note, as in all things I try to be empathetic and kind. I myself struggled greatly with these concepts as I transitioned into leadership, and I try and share personal anecdotes and my own struggles with the folks whom I support. Open your heart a little, it helps.</p>

<h1 id="scenarios">Scenarios</h1>
<ul>
  <li><a href="#badnews">Delivering Difficult News</a></li>
  <li><a href="#formerpeers">Loss of Peer Relationships</a></li>
  <li><a href="#trustjartruth">Assignment of Undesired Tasks</a></li>
  <li><a href="#opportunityleadership">Meeting Team Commitments while Growing a Team</a></li>
</ul>

<h2 id="badnews">Delivering Difficult News</h2>
<p><img alt="" src="/content/images/2022/badnews.jpeg" width="320" /></p>

<p>Here’s the scenario:</p>

<blockquote>
  <p>A leader is having difficulty delivering difficult/negative news to a team.</p>
</blockquote>

<p>Here’s the pitch:</p>

<blockquote>
  <p>By “protecting” your team from bad news, you create a culture of fear.</p>
</blockquote>

<p>Here’s the deal:</p>

<p>Early in my career, I thought that I was being a good leader by protecting my teams. I was causing them harm, but I didn’t realize it at the time. I thought I was being a good leader if I only gave a team the “good news”, the positive outlook, the cheery optimism of having the “best projects in the company.”  I thought I was giving them the proverbial “good jobs.”</p>

<p>What I was really doing was building an environment of existential fear. How?  By protecting these teams from the reality of the business and attempting to ‘smooth over’ company losses or poor results, I had eroded trust. Visibility, and real honesty drive trust within a team. In removing that, I had affected their psychological safety. They were always waiting for the other shoe to drop. Transparency, accountability, and revenue measurement are a form of kindness to a team. Trust isn’t just built on good news, as we’ll discuss in <a href="#trustjartruth">Assignment of Undesired Tasks</a> below.</p>

<h2 id="formerpeers">Loss of Peer Relationships</h2>
<p><img alt="" src="/content/images/2022/oppseverywhere.jpg" width="320" /></p>

<p>Here’s the scenario(s)  (this line of coaching is fairly universal):</p>

<blockquote>
  <ol>
    <li>A newly promoted leader is struggling with the loss of peer relationships on their team.</li>
    <li>A newly remote leader is struggling with building team culture over the internet.</li>
    <li>A leader is struggling to connect with a new team.</li>
  </ol>
</blockquote>

<p>Here’s the pitch:</p>

<blockquote>
  <p>Don’t be yourself, be the leader you want to be. Lead from your fundamentals, not your personality.</p>
</blockquote>

<p>Here’s the deal:</p>

<p>First, remind the manager that it’s their skillset that earned them this role, and it’s through that skillset and leadership experience that they can best help a team. Not personality. Not popularity.</p>

<p>I have found that a bold, clear statement of purpose helps to clear up any misconceptions. To wit, I recommend to all managers: Don’t be yourself!  Be the leader you want to be. Lead from your fundamentals, not your personality.</p>

<p>Jokes and interpersonal ribbing can ease tension, but possibly at the expense of someone’s emotional safety. There’s always an imbalance of power between a leader and those they support. Friendships are not built on power imbalances, and businesses do not run on friendships. All of the benefits of friendship in a business context flow upwards towards the business, and away from the individuals you support. “Oh, can you stay a few hours extra to help out <em>buddy?</em>”  “I know you have family commitments, <em>buuuut</em> can you come in early for a meeting just on Mondays?  To support your <em>work family</em>??”</p>

<p>Lead through your fundamentals. This is a kindness to those you support.</p>

<ol>
  <li>Build a culture of excellence and pair it with empathy.</li>
  <li>Set an expectation of mentorship and growth mindsets.</li>
  <li>Value accountability at all levels, have clear measures of success, share outcomes widely.</li>
  <li>Listen. If you find yourself interrupting folks, sip a drink during  1:1 meetings.</li>
</ol>

<h2 id="trustjartruth">Assignment of Undesired Tasks</h2>
<p><img alt="" src="/content/images/2022/trustjar.jpg" width="320" /></p>

<p>Here’s the scenario:</p>

<blockquote>
  <p>A leader is struggling with the assignment of undesirable tasks</p>
</blockquote>

<p>Here’s the pitch:</p>

<blockquote>
  <p>Trust isn’t built by throwing goodwill over a wall, it’s a relationship you build together collaboratively.</p>
</blockquote>

<p>Here’s the deal:</p>

<p>There’s this concept folks like to talk about in team leadership, that of the “trust jar”. It goes like this, every time you do something that builds trust with a team, you’re “putting a quarter into the trust jar.”  It’s a nice visual metaphor of the trust you build with a team over time.</p>

<p>The thing is, that’s only half the story. If you never have to ask for trust, if you never ask a team to suspend their disbelief and (in this case) do this undesirable task, then you’ve not built real trust. You’ve just banked goodwill. You’ve turned your trust jar into a wishing well, when it should really be more like the ‘take a penny leave a penny’ tray at the store. A shared trust, going both ways, building depth and value together. It is through this transparency and the shared shouldering of burdens that you truly build trust with your team.</p>

<h2 id="opportunityleadership">Meeting Team Commitments while Growing a Team</h2>
<p><img alt="" src="/content/images/2022/problemmountain.jpg" width="320" /></p>

<p>Here’s the scenario:</p>

<blockquote>
  <p>A leader is struggling with team growth while still meeting team commitments</p>
</blockquote>

<p>Here’s the pitch:</p>

<blockquote>
  <p>Every problem is an opportunity that someone needs, be an opportunity leader and pair them.</p>
</blockquote>

<p>Here’s the deal:</p>

<p>A business is really just a collection of problems. At all levels and at all times we’re facing challenges, prioritizing which ones to tackle and how to do so. An infinite mountain of problems, and we ski down the top slope of it every day. How then, to think about team growth when there are nothing but problems to face?</p>

<p>Remind those that you support of the mission. Of your responsibility as leaders to help others. Always tie it back to the mission, and to an empathetic outcome. Every problem that your business faces is really just an opportunity that someone in your organization needs. Maybe it’s the opportunity to learn a new skill. Maybe it’s the opportunity to help just one more person. Find that silver lining. Find the person who this challenge is also an opportunity for. As leaders we’re constantly measuring, learning, attempting to become better at what we do and in helping those we support. You should know (and if you don’t, well then get to knowing!) what areas your folks are looking to grow in. It’s a regular topic at your weekly check-ins, and you <em>never</em> want someone you support to be in the dark about their own performance and growth.  Pour yourself a nice slow-sippin drink, and get to <em>listening</em>.</p>

<p>Here’s a technique that I often employ with new leaders. I ask them if they know about the two purposes for every meeting. That usually piques their interest, and we dive in. The first purpose is obvious, it’s the stated purpose of the meeting. Every meeting has an outcome, decision or artifact that is clearly stated and obvious to all involved. If not, cancel that meeting as it’s an email in disguise.</p>

<p>The second purpose for every meeting is in opportunity leadership. Be on the lookout for problems that are really opportunities for folks you support (not just folks on your team, a good leader supports all departments.) You’d be surprised how quickly after finding out about a skill or growth area for an individual that an opportunity will present itself. Teaching this, truly embedding a growth mindset within your teams will pay off dividends for many generations of leadership. Encourage folks to talk about it plainly and transparently.</p>
]]></content:encoded>
        <pubDate>Mon, 10 Jan 2022 12:00:24 +0000</pubDate>
        <link>https://hunterdavis.com/2022/01/10/coaching-new-leaders-during-covid.html</link>
        <guid isPermaLink="true">https://hunterdavis.com/2022/01/10/coaching-new-leaders-during-covid.html</guid>
      </item>
    
  </channel>
</rss>
