That’s very nice. Feel free to integrate if you want :) Just request https://classify.stylometry.net/violations/49458161 or whatever item ID (as long as it was posted later than a few weeks ago) and it will return the scores for that post/comment and all its children.
Wow that was quick :) Meh, I feel like comment points have become somewhat untethered to comment quality (at least assuming threads are roughly sorted by comment score). It’s a good heuristic usually but it’s wrong often enough for it to be difficult to rely on.
I want to improve the classifier scores too. There’s some parameters in the loss function I could play around with more when training which basically optimize for more false negatives or false positives. Problems with no ground truth are very frustrating lol (though I suppose dang/tomhow’s manually written warnings could be treated as labels instead of using labels from GPT, but I suspect there’s not enough of them).
I'm pretty sure you would get into Berkeley with that profile. Also you can just major in math and get a job as a software engineer anyways. I agree that the system has reached the point of absurdity though. ETH Zurich and KU Leuven have the best solution to this issue in my mind, which is to let everyone in but make the classes ridiculously hard so that 50% leave in the first year.
> I'm pretty sure you would get into Berkeley with that profile
I was rejected from EECS@Cal and CS@CMU at the time (though my alum interviewer for CMU kept in contact with me to this day, and helped open doors for me professionally) yet got into an Ivy and waitlisted at MIT ¯\_(ツ)_/¯
And this was before college admissions got as competitive as it is for you guys today.
Most HNers graduated college in the 1990s and early 2000s and don't realize how competitive admissions have become across the board from 2008-09 onwards.
> Also you can just major in math and get a job as a software engineer anyways
Absolutely! Applied Math is a heavily interdisciplinary major that is very powerful.
Very interesting. Hunter S. Thompson described the same technique for learning to write. I'll try it even if it only gets applied to technical blog posts...
If you put important Anthropic blog posts like the Fable announcement or J-Space through Pangram, you get 100% human written. Considering that the overwhelming majority of the code there is written by AI, I think this is an admission that AI writing is slop and AI code is pretty good.
reCAPTCHA is a great success story of security through obscurity because probably less than 100 people have reverse engineered it and much less than that have produced a working solver for it that doesn't require a headless browser. Snapchat would be another good example - almost no one is going to put in the work to understand this [0]. Most companies just half ass it though and accordingly achieve nothing with the obscurity at all besides worse performance.
I wish we lived in a world where price wasn't the only way to filter out annoying passengers... if this experiment works maybe it will prove you can offer amenities more widely if you set high standards. I want an airline where at the end of the flight you rate the people next to you thumbs up or thumbs down and if someone gets >= 3 thumbs downs they are banned from the airline for 5 years. Eventually ordinary people would be able have the serene experience that only those in first class can presently enjoy as the people who get up thirty times to retrieve items from carry on or eat obnoxiously are removed.
I wish people would understand that extremist thinking is wrong and obtuse - just as it seems to most effective solution in your case it starts looking naive and damaging if you become the target just because you have a bad face or bad luck.
You are more than welcome to do so. Please keep in mind the realistic goal is raising the cost of scraping. Most bots use simple HTTP requests, and we make that useless.
