<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>falsify.dev — notes</title>
    <link>https://falsify.dev/notes</link>
    <description>Short essays on ML evaluation rigor, pre-registration, and the PRML specification.</description>
    <language>en-us</language>
    <atom:link href="https://falsify.dev/notes/feed.xml" rel="self" type="application/rss+xml" />
    <managingEditor>hello@falsify.dev (Cüneyt Öztürk)</managingEditor>
    <copyright>CC BY 4.0 — Cüneyt Öztürk</copyright>
    <lastBuildDate>Fri, 29 May 2026 12:00:00 GMT</lastBuildDate>
    <item>
      <title>Model cards vs pre-registration: what counts as evidence under the EU AI Act</title>
      <link>https://falsify.dev/notes/model-cards-vs-pre-registration</link>
      <guid isPermaLink="true">https://falsify.dev/notes/model-cards-vs-pre-registration</guid>
      <pubDate>Fri, 29 May 2026 12:00:00 GMT</pubDate>
      <author>hello@falsify.dev (Cüneyt Öztürk)</author>
      <description><![CDATA[The EU AI Act tells you to declare your accuracy numbers and keep records that let them be checked. It does not tell you which artifact satisfies that. A practical comparison of model cards and pre-registered manifests, and the one property that separates them: proof that the threshold was set before the run.]]></description>
    </item>
    <item>
      <title>Lock #2: the first thing PRML falsified was its own distribution hypothesis</title>
      <link>https://falsify.dev/notes/lock-2-postmortem</link>
      <guid isPermaLink="true">https://falsify.dev/notes/lock-2-postmortem</guid>
      <pubDate>Sat, 23 May 2026 00:30:00 GMT</pubDate>
      <author>hello@falsify.dev (Cüneyt Öztürk)</author>
      <description><![CDATA[Lock #2 resolved at 0 / 3 tonight. Zero independent contributors filed RFC engagement during the two-week window. The mechanism worked; the outreach didn't. Plus the counting bug caught two days before resolve, two awesome-list merges that did land, and what changes for v0.3.]]></description>
    </item>
    <item>
      <title>v0.2 RFC, briefly — what's open and why</title>
      <link>https://falsify.dev/notes/v0.2-rfc-open</link>
      <guid isPermaLink="true">https://falsify.dev/notes/v0.2-rfc-open</guid>
      <pubDate>Sun, 10 May 2026 12:00:00 GMT</pubDate>
      <author>hello@falsify.dev (Cüneyt Öztürk)</author>
      <description><![CDATA[Five proposals are open for community comment until 2026-05-22 23:59 UTC. v0.1 stays the stable spec. v0.2 will be additive — every v0.1 manifest hashes identically under v0.2 canonicalisation rules. The honest read of what changes, what doesn't, and why the comment window matters more than the proposals themselves.]]></description>
    </item>
  </channel>
</rss>
