<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: alxfrnr</title><link>https://news.ycombinator.com/user?id=alxfrnr</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Thu, 28 May 2026 19:18:05 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=alxfrnr" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by alxfrnr in "Investigating how prompt politeness affects LLM accuracy (2025)"]]></title><description><![CDATA[
<p>Dataset is way too small to be of any significance. It's just noise</p>
]]></description><pubDate>Thu, 28 May 2026 12:33:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=48308021</link><dc:creator>alxfrnr</dc:creator><comments>https://news.ycombinator.com/item?id=48308021</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48308021</guid></item><item><title><![CDATA[New comment by alxfrnr in "Files are the interface humans and agents interact with"]]></title><description><![CDATA[
<p>"They're portable, auditable, and composable. No MCP server required. No plugin marketplace to browse. Just a folder with a SKILL.md in it." It screams lazy ai slop copy-paste. Now when I see this kind of obvious llm tell, I can't help but feel betrayed for investing time in giving it a shot in the first place.</p>
]]></description><pubDate>Sun, 08 Mar 2026 12:00:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=47296642</link><dc:creator>alxfrnr</dc:creator><comments>https://news.ycombinator.com/item?id=47296642</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47296642</guid></item><item><title><![CDATA[New comment by alxfrnr in "Instant Checkout and the Agentic Commerce Protocol"]]></title><description><![CDATA[
<p>So they use all the reviews and product testing they ingest to come-up with results, then they let you buy whatever they list in 1-click and bank a commission. A comission that would have gone in the pockets of an affiliate with actual hands on experience on the product (not talking about all the SEO crap of course), effectively killing the affiliates who provided the needed data for the rankings in the first place.<p>It also paves the way for ads in this type of answer down the line and incentivizes enshitification even further. If it would be Google doing it, everybody would raise their shields. But as openAI is still too small, we'll let it happen without saying or doing anything, right?<p>It hides stores/brands reputation, value-prop and differenciators, likely favoring the established big players and makes it even harder for smaller players to enter.<p>What about the ethics of having openAI as the gatekeeper for product/brands inclusion? Right now it mighht look like a clever business move but it establishes dangerous foundations and gives them too much power if they keep growing.</p>
]]></description><pubDate>Tue, 30 Sep 2025 10:42:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=45423949</link><dc:creator>alxfrnr</dc:creator><comments>https://news.ycombinator.com/item?id=45423949</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45423949</guid></item></channel></rss>