alt Hacker NewsShow HN: I Built Paul Graham's Intellectual Captcha Idea
PG has posted about improving social networks using something like an "intellectual CAPTCHA" many times [1][2][3][4] - "Make users pass a test on basic concepts like the distinction between necessary and sufficient conditions before they can tweet."
I felt the same way. So I built one using a mix of simple math, logic, and Twitter/X Community Noted posts. Try sample questions here - https://mentwire.com/sample - without signing up.
- Invites are temporarily open to HN users.
- Onboarding test + one daily question before accessing feed, post or reply.
- Posts authors are anonymous until upvoted or downvoted, forcing evaluation of content on merit.
- Face ID (on-device only) to post/reply, pangram checks for AI text.
Sourcing good questions turned out to be much harder than I thought. If you have suggestions to scale this, I would love to hear. Eventually, could be gated across disciplines/topics to get a competence × interest graph instead of the pure interest graph of today's social networks.
[1] https://x.com/paulg/status/1235949761359904768 [2] https://x.com/paulg/status/1576517990182359040 [3] https://x.com/paulg/status/1514979883948126209 [4] https://x.com/paulg/status/1505842647319126016
Comments
One issue with this is that it mixes hypothetical formal logic style problems (where there are clear, inflexible rules) with real life examples (where group membership/traits, cost estimation, and causal attribution are less clear) without always disambiguating which one is which. Fun quiz though!
I opened it, it told me it was impossible to build a house in california for less than 350K, i closed it
This is weird political propaganda. The first post misrepresented annual costs of housing.
Two mild concerns: first, I missed one and it told me I didn’t miss any at the end.
Second, some of the logic problems have flawed premises (eg All licensed pilots must pass a medical exam. Jake is a licensed pilot, therefore Jake passed a medical exam.) If you see the flaw in the premise (it assumes no fraud) then the conclusion does not follow.
Im not sure you’re going to be able to actually improve human discourse this way. The idea that it’s ‘irrationality’ that’s the source of xitters problems is far too shallow to really make a change.
I answered 8/10 correctly but mostly on instinct, for example betting that the Trump tweet is misleading. Opus 4.6 got 9/10 correct. You might need an internal time limit (don't show the user) and some strawberry questions.
Reminds me of IQ tests I took as a kid.
"Finish the sequence" with 4 options and "no pattern" as the choices.
It becomes "what does the moderately intelligent person who wrote the test thinks counts as a pattern" not the intended exercise at all. There was never enough samples to even guess at a real pattern in them.
> "Make users pass a test on basic concepts like the distinction between necessary and sufficient conditions before they can tweet."
If twitter ever became what he says he wants, he'd quit using it within a month. He already has the option to close twitter and seek out experts' writing. Why is he choosing to bask in the emotions generated by people being wrong on twitter?
It's like listening to a friend complain about twitter being "full of" content that you rarely/never see on your feed. Nah, that's their algorithm and they just told you exactly who they are.
Funny thing had to laugh :)
Intellectual?
it's a really nice idea, but of course completely antithetical to the business model of modern social media platforms. So, it will never go anywhere. HN might be the only locale with any real numbers that I could see actually using it. Even BlueSky I think could never risk something like this.
as an interesting thought experiment, consider the questions that TruthSocial would put in. would an average unsophisticated user be able to tell the difference between your product and a hopelessly biased version such as that? they would support the correct answers with their own misinformation. Would it be just another schism of reality?
At least one of the test questions was just a screen shot from a tweet. It was difficult to read. I'd suggest extracting text from screen shots with OCR. Apple has built-in functionality for this on their operating systems with Live Text. There are strong open source systems based on small vision language models for this, too. The one I have been recommending lately is GLM-OCR:
https://github.com/zai-org/GLM-OCR
It's fast and can run even on low-resource computers.
---
Does this CAPTCHA actually resist computers? I didn't try feeding the questions I got to an LLM, but my sense is that current frontier models could probably pass all of these too. Making generated text pass the pangram test is simple enough for someone actually writing a bot to spin up automated accounts.