Hacker News: davidajackson

Show HN: Loophole – a word game where you make the AI say the secret word

davidajackson — Wed, 22 Jul 2026 16:58:13 +0000

Hi HN. I made a word game where you get a secret word and a list of banned words, and you write clues to make an AI say the secret. One word per day, same for everyone.

The first version was solve/fail and it was boring, because the guesser solves basically any coherent clue in one shot. I gave it "cyclops of the harbor" and it immediately said lighthouse. So scoring works like golf now: each word has a par, which I get by running a pile of plausible human clues against the pinned model and taking the median, and you're trying to beat par on total word count. Annoying side effect: pars are relative to that specific model, so if I ever upgrade it I have to recalibrate everything.

Cheating was the harder problem. Banned words are checked deterministically with lemma matching, but that does nothing against "mariposa" for butterfly, or "vol-KAY-no", or rhymes. I wrote a red-team script and it found ~99 automatic wins like that across 37 words. There's now a second model call that runs in parallel with the guesser (so no added latency) and rejects translation/phonetic/rhyme/spelling clues. Clues that work through meaning are still legal — "the thing Vesuvius did" is fine, which is sort of the point of the name.

The secret word never leaves the server until the game ends, so a rejected clue can't be used as an oracle to narrow it down.

Honestly the appeal ended up being the guesser's failures. It's dead serious and confidently wrong a lot ;)

Free daily, no account, no ads. iOS only for now. If you find a clue style that beats the judge I want to hear about it.

Comments URL: https://news.ycombinator.com/item?id=49009833

Points: 2

# Comments: 0

Ask HN: Illustration Themes in AI Apps?

davidajackson — Fri, 10 Jul 2026 14:19:50 +0000

Have any of you been successful in generating authentic drawing styles via AI apps? Drawing design/scribble/actual good looking hand drawings.

Comments URL: https://news.ycombinator.com/item?id=48860327

Points: 2

# Comments: 0

New comment by davidajackson in "Ask HN: Does Claude Code remove the need for so many front-end frameworks?"

davidajackson — Sun, 31 May 2026 18:02:33 +0000

Yes. Memorizing syntax is now obsolete. Frameworks are generally somewhat constrained, now they can be auto-generated to be exactly the UI you want. Only caveat would be saving tokens but tokens are cheap. RIP

Ask HN: Agentic Prompt Compaction Strategies

davidajackson — Fri, 24 Apr 2026 14:45:13 +0000

What are your favorite reasoning/compaction strategies for saving token spend, and why?

Comments URL: https://news.ycombinator.com/item?id=47891020

Points: 1

# Comments: 0

New comment by davidajackson in "Ask HN: Am I getting old, or is working with AI juniors becoming a nightmare?"

davidajackson — Fri, 24 Apr 2026 14:43:24 +0000

Many here will be sad but there will be a day when writing code is seen as as antiquated as using a slide rule. It is coming.

Ask HN: What function will inference cost take v.s. time?

davidajackson — Thu, 16 Apr 2026 13:12:03 +0000

If you had to project what kind of pattern we'll see in inference cost v.s. time, what would you predict? Over the next 2, 5, 10+ years?

Comments URL: https://news.ycombinator.com/item?id=47792471

Points: 2

# Comments: 0

Ask HN: Infinitely Scalable Prompts

davidajackson — Mon, 13 Apr 2026 14:04:13 +0000

Prompts and prompt rot could potentially be solved by dynamic injection of relevant, embedded snippets to dynamically build prompts. Right now most company's prompts are too small to suffer from such bloat, but I'm guessing soon we'll see it.

Have any of you tried this? What was the LOE and did you benchmark it? What did you find?

Comments URL: https://news.ycombinator.com/item?id=47752156

Points: 2

# Comments: 0

Ask HN: Agentic Permutation of Testing Paths In A System

davidajackson — Sat, 11 Apr 2026 20:34:38 +0000

Given a system with N components and M pairwise connections between said N components, there exists infinite sequences of [component, command, ...] for a series of operations. A component displays data and a command obviously executes a command.

Has anyone here done research on using an agant to select happy paths to test based on user intent? Is there literature on this? Seems like a cost, compute, intent optimization problem.

Assume latency is a restriction.

Seems like something some AI research folks here can chime in on. I have my own opinions but want to hear what the community thinks.

Comments URL: https://news.ycombinator.com/item?id=47733790

Points: 4

# Comments: 0

Ask HN: Neuromorphic Computing

davidajackson — Wed, 11 Mar 2026 18:35:33 +0000

Does anyone here have a lot of knowledge on neuromorphic computing, or has worked in the space? What do you think of it? Also, do you see any ethical issues with this type of engineering...

Comments URL: https://news.ycombinator.com/item?id=47339401

Points: 2

# Comments: 0

New comment by davidajackson in "Ask HN: What is still hard about system design with AI?"

davidajackson — Mon, 22 Dec 2025 21:06:23 +0000

That's a people problem, not an AI problem. People spend so much time debating overcomplicating early system design (API v.s. event driven, etc.). AI is pretty good at generating what you want if you tell it your LOE and resources.

New comment by davidajackson in "Ask HN: Why does Y Combinator seem to be consistently funding AI slop?"

davidajackson — Wed, 19 Nov 2025 16:28:44 +0000

I am going to play devil's advocate here and say that so many processes that can be automated with AI are still untouched.

I'm not making a statement about the quality of businesses funded, but rather the fact there is a lot of opportunity.

Ask HN: Image Recognition

davidajackson — Wed, 19 Nov 2025 16:26:58 +0000

In this thread I would like to discuss the interesting ways we've found to increase model accuracy when analyzing images. I have been learning about vision models and trying a few different strategies, to varying degrees of success, and would like to brainstorm a bit further.

To that end, I have a few questions for HN:

What's your experience with bounding boxes?

What's your experience with different models/what are your favorites?

What's your experience with triaging extraction errors while managing cost?

Comments URL: https://news.ycombinator.com/item?id=45981504

Points: 2

# Comments: 0

New comment by davidajackson in "Ask HN: Is A2A the Future?"

davidajackson — Mon, 22 Sep 2025 02:42:56 +0000

sorry, agent to agent protocol or agent to env

https://developers.googleblog.com/en/a2a-a-new-era-of-agent-...

Ask HN: Is A2A the Future?

davidajackson — Sun, 21 Sep 2025 14:50:00 +0000

Wanted to have a brief discussion here as to whether you see A2A as the 'final state' for most AI automation. Are there any alternative patterns that you feel might be competitive, or better than A2A? Thoughts on A2E?

Comments URL: https://news.ycombinator.com/item?id=45323255

Points: 1

# Comments: 3

Ask HN: Behavioral assumptions in self driving cars

davidajackson — Tue, 16 Sep 2025 15:16:51 +0000

Do self driving cars ever make assumptions about the lack of desirability of a lane if there is no one in it? While counterintuitive, sometimes that is an indication that there might be a parked car not currently visible etc. This has struck me as a curious problem where cars would actually observe other car's decisions ahead to guess that 'something is up'.

Comments URL: https://news.ycombinator.com/item?id=45263471

Points: 1

# Comments: 0

New comment by davidajackson in "graduated but no jobs"

davidajackson — Sun, 14 Sep 2025 02:04:45 +0000

Do you have a portfolio site of sample projects or case studies? Do you want us to critique that?

New comment by davidajackson in "Ask HN: Theory That Economic Growth Stagnates in a Civilization"

davidajackson — Fri, 29 Aug 2025 16:42:22 +0000

No doubt about the zero sum game in terms of a possible path in terms of wealth distribution. My question is what will be the function of the "sum part" as a society builds all technologies to 'completion' which is defined as when the utility of any new technology approaches zero because there is nothing 'valuable' or 'new to invent'.

A different argument is that as technology improves to infinity, wealth also becomes ubiquitous and potentially infinite. So even though no new technology is invented that is valuable, there is infinite wealth through abundance.

It seems to me that the primary driver of which world we end up in is whether humanity is able to become multi-planetary. Otherwise, resources become a constraint and instead of discovering new existing resources, humanity can only create value by inventing new technology.

These are not strong opinions.

Ask HN: Theory That Economic Growth Stagnates in a Civilization

davidajackson — Fri, 29 Aug 2025 15:01:35 +0000

I have had a theory for a while that economic growth or the economy in terms of CAGR will go to 0% or becomes linear with population as the population grows. This in my mind aligns with the opportunities for lifestyle improvements going to 0 (we already have the internet, we already have food, shelter etc.) so this is no new market to cover/no new businesses to invent?

Has this topic been discussed or debated anywhere? Has anyone with this same theory done any mathematical projections?

Comments URL: https://news.ycombinator.com/item?id=45065049

Points: 4

# Comments: 4

New comment by davidajackson in "Ask HN: How to teach a 4 year old to code?"

davidajackson — Fri, 29 Aug 2025 14:51:37 +0000

Make it a game like the carol the robot class at stanford.

Ask HN: Is any logic instinctive, or all trained?

davidajackson — Fri, 29 Aug 2025 14:49:51 +0000

Is there evidence that understanding causality (action -> reaction, cause -> effect, etc.) is instinctive or does it need to be trained in humans? It seems obvious that humans need to be taught some forms of logic, like logical fallacies, but are there some that are instinctive? How has this been 'proved'?

Comments URL: https://news.ycombinator.com/item?id=45064890

Points: 2

# Comments: 0