<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Communication on Brenton Mallen</title><link>https://brentonmallen.com/tags/communication/</link><description>Recent content in Communication on Brenton Mallen</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Wed, 22 Jul 2026 12:00:00 -0400</lastBuildDate><atom:link href="https://brentonmallen.com/tags/communication/index.xml" rel="self" type="application/rss+xml"/><item><title>The Weight of "Just"</title><link>https://brentonmallen.com/posts/the-weight-of-just/</link><pubDate>Wed, 22 Jul 2026 12:00:00 -0400</pubDate><guid>https://brentonmallen.com/posts/the-weight-of-just/</guid><description>&lt;p>&lt;em>&amp;ldquo;Can&amp;rsquo;t you just retrain the model?&amp;rdquo;&lt;/em>&lt;/p>
&lt;p>I&amp;rsquo;ve heard some version of this sentence more times than I can count, and if I&amp;rsquo;m
being honest, I&amp;rsquo;ve said my fair share of them too. It sounds harmless. Its
usage has become so common that its presence seems to go unnoticed, but it&amp;rsquo;s
already doing damage, because under that &amp;ldquo;just&amp;rdquo;
sits an iceberg: figuring out whether the data drifted or the labels did,
sourcing new labels, rebuilding the feature pipeline that quietly broke two
months ago, validating against the old model, regression testing, and the
deployment, all while wrestling the compounding interest of tech debt. This is really the
three sprints that &amp;ldquo;retrain&amp;rdquo; actually costs.&lt;/p></description></item></channel></rss>