<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: kmad</title><link>https://news.ycombinator.com/user?id=kmad</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Fri, 17 Jul 2026 04:18:15 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=kmad" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[Token Capital Efficiency]]></title><description><![CDATA[
<p>Article URL: <a href="https://kmad.ai/Token-Capital-Efficiency">https://kmad.ai/Token-Capital-Efficiency</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=48711616">https://news.ycombinator.com/item?id=48711616</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Sun, 28 Jun 2026 21:01:59 +0000</pubDate><link>https://kmad.ai/Token-Capital-Efficiency</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=48711616</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48711616</guid></item><item><title><![CDATA[New comment by kmad in "Agents need control flow, not more prompts"]]></title><description><![CDATA[
<p>This is, at least in part, the promise of frameworks like DSPy and PydanticAI. They allow you to structure LLM calls within the broader control flow of the program, with typed inputs and outputs. That doesn’t fix non-determinism, hallucinations, etc., but it does allow you to decompose what it is you’re trying to accomplish and be very precise about when an LLM is called and why.</p>
]]></description><pubDate>Thu, 07 May 2026 19:17:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=48053572</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=48053572</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48053572</guid></item><item><title><![CDATA[Achieving 20%+ improvement in structured extraction using DSPy and GEPA]]></title><description><![CDATA[
<p>Article URL: <a href="https://kmad.ai/DSPy-Optimization">https://kmad.ai/DSPy-Optimization</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46289313">https://news.ycombinator.com/item?id=46289313</a></p>
<p>Points: 3</p>
<p># Comments: 1</p>
]]></description><pubDate>Tue, 16 Dec 2025 14:58:41 +0000</pubDate><link>https://kmad.ai/DSPy-Optimization</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=46289313</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46289313</guid></item><item><title><![CDATA[Notes On: Measuring AI Ability to Complete Long Tasks]]></title><description><![CDATA[
<p>Article URL: <a href="https://kmad.ai/AI-Long-Horizon-Tasks">https://kmad.ai/AI-Long-Horizon-Tasks</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=43821295">https://news.ycombinator.com/item?id=43821295</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 28 Apr 2025 13:25:22 +0000</pubDate><link>https://kmad.ai/AI-Long-Horizon-Tasks</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=43821295</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43821295</guid></item><item><title><![CDATA[New comment by kmad in "Ask HN: What are you working on? (March 2025)"]]></title><description><![CDATA[
<p>Awesome! Have you considered pyodide[1]? Pydantic uses this for sandboxing its AI agents [2].<p>1. <a href="https://pyodide.org/en/stable/" rel="nofollow">https://pyodide.org/en/stable/</a>
2. <a href="https://ai.pydantic.dev/mcp/run-python/" rel="nofollow">https://ai.pydantic.dev/mcp/run-python/</a></p>
]]></description><pubDate>Sun, 30 Mar 2025 23:34:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=43528907</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=43528907</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43528907</guid></item><item><title><![CDATA[New comment by kmad in "An LLM plugin to support Structured Outputs"]]></title><description><![CDATA[
<p>A plugin for Simon Willison's llm tool to support Structured Outputs using a simple YAML definition file.</p>
]]></description><pubDate>Tue, 14 Jan 2025 14:14:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=42697429</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=42697429</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42697429</guid></item><item><title><![CDATA[An LLM plugin to support Structured Outputs]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/kmad/llm-structure">https://github.com/kmad/llm-structure</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=42697428">https://news.ycombinator.com/item?id=42697428</a></p>
<p>Points: 1</p>
<p># Comments: 1</p>
]]></description><pubDate>Tue, 14 Jan 2025 14:14:50 +0000</pubDate><link>https://github.com/kmad/llm-structure</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=42697428</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42697428</guid></item><item><title><![CDATA[Measuring LLM Confidence with Structured Outputs]]></title><description><![CDATA[
<p>Article URL: <a href="https://kmad0.github.io/Measuring-LLM-Confidence">https://kmad0.github.io/Measuring-LLM-Confidence</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=42467581">https://news.ycombinator.com/item?id=42467581</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 20 Dec 2024 02:00:43 +0000</pubDate><link>https://kmad0.github.io/Measuring-LLM-Confidence</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=42467581</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42467581</guid></item><item><title><![CDATA[New comment by kmad in "Fiscal Dominance and the Return of Zero-Interest Bank Reserve Requirements [pdf]"]]></title><description><![CDATA[
<p>Abstract:<p>As a matter of arithmetic, the trends of US government debt and deficits will eventually result in an outrageously high government debt-to-GDP ratio. But when exactly will the United States hit the constraint of infeasibility and how exactly will policy adjust to it? This article considers fiscal dominance, which is the possibility that accumulating government debt and deficits can produce increases in inflation that “dominate” central bank intentions to keep inflation low. Is it a serious possibility for the United States in the near future?
And how might various policies change (especially those related to the banking system) if fiscal dominance became a reality?</p>
]]></description><pubDate>Wed, 19 Jul 2023 02:42:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=36781530</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=36781530</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36781530</guid></item><item><title><![CDATA[Fiscal Dominance and the Return of Zero-Interest Bank Reserve Requirements [pdf]]]></title><description><![CDATA[
<p>Article URL: <a href="https://files.stlouisfed.org/files/htdocs/publications/review/2023/06/02/fiscal-dominance-and-the-return-of-zero-interest-bank-reserve-requirements.pdf">https://files.stlouisfed.org/files/htdocs/publications/review/2023/06/02/fiscal-dominance-and-the-return-of-zero-interest-bank-reserve-requirements.pdf</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=36781529">https://news.ycombinator.com/item?id=36781529</a></p>
<p>Points: 4</p>
<p># Comments: 1</p>
]]></description><pubDate>Wed, 19 Jul 2023 02:42:51 +0000</pubDate><link>https://files.stlouisfed.org/files/htdocs/publications/review/2023/06/02/fiscal-dominance-and-the-return-of-zero-interest-bank-reserve-requirements.pdf</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=36781529</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36781529</guid></item><item><title><![CDATA[New comment by kmad in "Matrices and Graph"]]></title><description><![CDATA[
<p>This approach reminds me of RedisGraph[1] (which is now unfortunately EoL).<p>"RedisGraph is the first queryable Property Graph database to use sparse matrices to represent the adjacency matrix in graphs and linear algebra to query the graph."<p>1. <a href="https://github.com/RedisGraph/RedisGraph">https://github.com/RedisGraph/RedisGraph</a></p>
]]></description><pubDate>Sat, 15 Jul 2023 14:04:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=36736862</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=36736862</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36736862</guid></item><item><title><![CDATA[New comment by kmad in "E3: Microsoft’s next Xbox: 8K graphics, SSD storage, and ray-tracing for 2020"]]></title><description><![CDATA[
<p>Sounds like a perfect application for Micron/Intel's (yet to be widely commercialized) 3D XPoint[1] :) Although I find it highly unlikely a new Xbox would be the place to debut the technology.<p>Regardless, claimed latency is between DRAM and NAND, so I wonder if it would be performant enough for a gaming use case. To mikeash's point, 3D XPoint is byte-addressable, so perhaps it's possible...<p>[1] <a href="https://www.micron.com/products/advanced-solutions/3d-xpoint-technology" rel="nofollow">https://www.micron.com/products/advanced-solutions/3d-xpoint...</a></p>
]]></description><pubDate>Sun, 09 Jun 2019 23:48:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=20142659</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=20142659</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=20142659</guid></item><item><title><![CDATA[New comment by kmad in "Why I don't like smartcards, HSMs, YubiKeys, etc."]]></title><description><![CDATA[
<p>How does something like the U2F Zero[1] compare?<p>As I understand it, the u2f zero acts as an HID device and not as a smartcard provider, but could one modify the firmware to do that? Isn't this basically an open source yubikey you can make yourself for < $25?<p>1. <a href="https://github.com/conorpp/u2f-zero" rel="nofollow">https://github.com/conorpp/u2f-zero</a></p>
]]></description><pubDate>Thu, 24 Nov 2016 19:48:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=13032347</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=13032347</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=13032347</guid></item><item><title><![CDATA[New comment by kmad in "Adblock Browser for Android"]]></title><description><![CDATA[
<p>Just going off memory, but something like this seems to work for me on rooted Android (or wherever your hosts file is):<p>curl <a href="http://someonewhocares.org/hosts/hosts" rel="nofollow">http://someonewhocares.org/hosts/hosts</a> > /system/etc/hosts<p>Of course all the normal disclaimer stuff, make sure you look at what you're putting in there, etc. But Mr. Pollock's list has worked wonders for me</p>
]]></description><pubDate>Wed, 20 May 2015 14:47:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=9576664</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=9576664</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=9576664</guid></item><item><title><![CDATA[New comment by kmad in "Ask HN: You want Atom invites/you have invites to give?"]]></title><description><![CDATA[
<p>Would love an invite! - kmadura AT googlemail</p>
]]></description><pubDate>Thu, 27 Feb 2014 15:01:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=7312496</link><dc:creator>kmad</dc:creator><comments>https://news.ycombinator.com/item?id=7312496</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=7312496</guid></item></channel></rss>