https://alexcbecker.net/blog.html Blog posts by Alex Becker en-us https://alexcbecker.net/blog/inventing-transformers.html https://alexcbecker.net/blog/inventing-transformers.html Sat, 30 May 2026 00:00:00 +0000 A branching tech tree of the Transformer: the path of innovations from 2012 to 2017 - AlexNet, Word2vec, Attention, ResNet and more - drawn as glowing nodes on a skill-tree diagram, each one unlocking the next on the way to the architecture behind modern AI. AI History https://alexcbecker.net/blog/myth-of-deterministic-software.html https://alexcbecker.net/blog/myth-of-deterministic-software.html Sun, 03 May 2026 00:00:00 +0000 How I've come to terms with LLM non-determinism in software and abandoned the comforting lies we've been telling ourselves about traditional software. AI Software https://alexcbecker.net/blog/decreasing-surplus.html https://alexcbecker.net/blog/decreasing-surplus.html Sat, 25 Apr 2026 00:00:00 +0000 The feeling of poverty amidst wealth stems from less opportunity to realize surplus. Bespoke goods are bid up, and market opportunities narrow for consumers & producers. Economics https://alexcbecker.net/blog/rationalisms-lamprey.html https://alexcbecker.net/blog/rationalisms-lamprey.html Wed, 22 Apr 2026 00:00:00 +0000 How Steve Sailer wormed his way into Silicon Valley's hivemind through Slate Star Codex. Rationalism Politics https://alexcbecker.net/blog/how-claude-changes-coding.html https://alexcbecker.net/blog/how-claude-changes-coding.html Mon, 06 Apr 2026 00:00:00 +0000 Claude writes all my code now, but the bigger shift is in *what* code I produce. AI Software https://alexcbecker.net/blog/prompt-injection-before-anthropic.html https://alexcbecker.net/blog/prompt-injection-before-anthropic.html Mon, 23 Feb 2026 00:00:00 +0000 Prompt injection requires training LLMs with robust "untrusted content" primitives, viewing it as a solvable cybersecurity problem distinct from jailbreaking. AI Cybersecurity https://alexcbecker.net/blog/startup-equity.html https://alexcbecker.net/blog/startup-equity.html Wed, 11 Feb 2026 00:00:00 +0000 It is in startups' interest to pay employees cash and save their equity to sell to VCs. Startups https://alexcbecker.net/blog/boldness-alone-is-no-virtue.html https://alexcbecker.net/blog/boldness-alone-is-no-virtue.html Wed, 07 Jan 2026 00:00:00 +0000 Silicon Valley backs Trump for "bold action," because America's problems require it. But why aren't his bold actions aimed at addressing those problems? Politics https://alexcbecker.net/blog/why-everyone-feels-poor.html https://alexcbecker.net/blog/why-everyone-feels-poor.html Sat, 03 Jan 2026 00:00:00 +0000 Why do rich Americans feel poor? Is inflation understated, or are feelings driven by relative wealth, materialism, the rise of unattainable luxury goods, and a crisis of meaning? Economics https://alexcbecker.net/blog/prompt-injection-benchmark.html https://alexcbecker.net/blog/prompt-injection-benchmark.html Thu, 30 Oct 2025 00:00:00 +0000 I built a new prompt injection benchmark for AI agents because the existing ones are not great. AI Cybersecurity https://alexcbecker.net/blog/usenix-security-25.html https://alexcbecker.net/blog/usenix-security-25.html Sat, 16 Aug 2025 00:00:00 +0000 USENIX Security '25 roundup on LLM security topics: prompt injection, data leaks, CAPTCHA bypasses, and model misbehavior. AI Cybersecurity https://alexcbecker.net/blog/chinese-room.html https://alexcbecker.net/blog/chinese-room.html Wed, 25 Jun 2025 00:00:00 +0000 The Chinese Room argument is flawed because it ignores the practical speed and scale of computation, misguiding our intuition about AI intelligence. AI Philosophy https://alexcbecker.net/blog/prompt-injection.html https://alexcbecker.net/blog/prompt-injection.html Sat, 21 Jun 2025 00:00:00 +0000 Prompt injection, where untrusted input overrides LLM system instructions, is a hard problem with no reliable defenses yet. AI Cybersecurity https://alexcbecker.net/blog/ai-learning-theory.html https://alexcbecker.net/blog/ai-learning-theory.html Sun, 30 Mar 2025 00:00:00 +0000 LLMs display "generalized intelligence," yet research lacks a theoretical explanation for their capabilities, how they work, or their inherent limitations, and there is no clear path to fixing this. AI https://alexcbecker.net/blog/transformers.html https://alexcbecker.net/blog/transformers.html Sat, 22 Mar 2025 00:00:00 +0000 Writing stuff down about the Transformer architecture and training methods to solidify my own understanding. AI https://alexcbecker.net/blog/ai-and-the-future-economy.html https://alexcbecker.net/blog/ai-and-the-future-economy.html Sat, 22 Mar 2025 00:00:00 +0000 AI's economic future is debated. I explore 5 scenarios based on differing AI cognitive & robotic progress assumptions. AI Economics https://alexcbecker.net/blog/democrats-lost-whats-next.html https://alexcbecker.net/blog/democrats-lost-whats-next.html Sun, 10 Nov 2024 00:00:00 +0000 Democrats lost due to inflation & immigration. To win again, they need stronger central leadership, to shed toxic stances, and to prioritize pragmatic solutions on key issues. Politics https://alexcbecker.net/blog/israel-palestine.html https://alexcbecker.net/blog/israel-palestine.html Sun, 03 Nov 2024 00:00:00 +0000 Israel holds nearly all power in the conflict. A two-state solution is the most realistic and moral path, primarily requiring Israeli action despite limited external influence. Politics https://alexcbecker.net/blog/satoshis-legacy.html https://alexcbecker.net/blog/satoshis-legacy.html Mon, 21 Oct 2024 00:00:00 +0000 Satoshi's Bitcoin, intended for freedom, has failed to realize Finney's ambitions and instead empowered global crime, scams, and human trafficking. Cryptocurrency Economics Ethics https://alexcbecker.net/blog/ai-and-the-future-of-work.html https://alexcbecker.net/blog/ai-and-the-future-of-work.html Sun, 20 Oct 2024 00:00:00 +0000 AI will transform work, replacing some jobs & enhancing others, but its crucial impact lies in the "signaling problem," where AI-generated content devalues human competence. AI Economics https://alexcbecker.net/blog/a-machine-for-burning-cities.html https://alexcbecker.net/blog/a-machine-for-burning-cities.html Sun, 24 Mar 2019 00:00:00 +0000 "Nuclear" bombs aren't all equal. Fusion weapons are exponentially more destructive than fission bombs—a vital distinction missed in public understanding of global threats. Risk History https://alexcbecker.net/blog/user-projections.html https://alexcbecker.net/blog/user-projections.html Sun, 02 Dec 2018 00:00:00 +0000 How to project SaaS user growth? This post shows why simple median/mean models of monthly user *change* are often best. Data Science Statistics https://alexcbecker.net/blog/in-blockchain-we-trust.html https://alexcbecker.net/blog/in-blockchain-we-trust.html Mon, 04 Jun 2018 00:00:00 +0000 Blockchains promise "trustless" transactions, but they actually require new forms of trust in value and network security, making them vulnerable to major attacks. Cryptocurrency Economics https://alexcbecker.net/blog/electronic-health-records-considered-harmful.html https://alexcbecker.net/blog/electronic-health-records-considered-harmful.html Mon, 05 Feb 2018 00:00:00 +0000 Electronic Health Records (EHRs) harm patient care & raise costs due to bad design, lack of interoperability, and government mandates. Healthcare Software Policy https://alexcbecker.net/blog/a-front-end-debugging-adventure.html https://alexcbecker.net/blog/a-front-end-debugging-adventure.html Sun, 31 Dec 2017 00:00:00 +0000 Debugging a front-end Firefox bug: A double-click race condition caused `mousedown` to overwrite an element's reference before `mouseup` could remove it. Software https://alexcbecker.net/blog/what-mathematics-has-lost.html https://alexcbecker.net/blog/what-mathematics-has-lost.html Sun, 03 Sep 2017 00:00:00 +0000 Mathematicians often fail to share the human intuitions behind their work, leading to a profound loss of understanding, exemplified by Mirzakhani's death. Mathematics https://alexcbecker.net/blog/please-lie.html https://alexcbecker.net/blog/please-lie.html Thu, 13 Jul 2017 00:00:00 +0000 Fight back against pervasive surveillance & exploitative algorithms: lie about your data to disrupt models that stereotype, entrench poverty, and reinforce corporate power. Ethics https://alexcbecker.net/blog/the-cult-of-averages.html https://alexcbecker.net/blog/the-cult-of-averages.html Sun, 25 Jun 2017 00:00:00 +0000 We often think in terms of the mean even when this is ill-defined. Medians are usually better, but there is no substitute for looking at the full distribution. Statistics https://alexcbecker.net/blog/lessons-from-milo.html https://alexcbecker.net/blog/lessons-from-milo.html Sat, 11 Mar 2017 00:00:00 +0000 Violent protests against Milo Yiannopoulos ironically did his work for him. Politics https://alexcbecker.net/blog/against-ai-risk.html https://alexcbecker.net/blog/against-ai-risk.html Sun, 27 Nov 2016 00:00:00 +0000 Grey goo scenarios obscure the more imminent and likely risks: military AI, inequality-fueling tech, and other powerful tools society already struggles to control. AI Rationalism https://alexcbecker.net/blog/you-are-probably-misusing-opacity.html https://alexcbecker.net/blog/you-are-probably-misusing-opacity.html Sat, 29 Oct 2016 00:00:00 +0000 Designers often misuse opacity for text, leading to unreadable stacking effects and CSS bugs like new stacking contexts. Use fixed colors instead! Software https://alexcbecker.net/blog/analyzing-the-effects-of-voting.html https://alexcbecker.net/blog/analyzing-the-effects-of-voting.html Sat, 29 Oct 2016 00:00:00 +0000 An individual vote has never tipped an important election, so why vote? Politics Statistics https://alexcbecker.net/blog/towards-universal-https-adoption.html https://alexcbecker.net/blog/towards-universal-https-adoption.html Sun, 24 Jul 2016 00:00:00 +0000 Why is HTTPS adoption so slow? Strict browser mixed-content policies. A "No-Security-Claim" header could simplify partial security & boost web encryption. Software https://alexcbecker.net/blog/django-is-dangerous.html https://alexcbecker.net/blog/django-is-dangerous.html Mon, 06 Jun 2016 00:00:00 +0000 This post argues Django is dangerous due to fundamental flaws in data integrity and dynamic templating that worsen as projects scale. Software