@aaronberdofe Like, in Roslindale, my house was built with 4-5 ft side setback, but now it’s 10. Why?
Was it a false correlation of “density brings crime” or something like that out of the white flight/urban decay era?
@aaronberdofe Is there anything written on “what was going on when all of this more restrictive zoning (from a setback/density perspective) was passed”?
I wish there was a documentary where they interviewed people from that time pushing these policies. What did they want to accomplish?
@WiseWyzard@KathrynTewson Or is even the lawsuit/serving process very difficult/costly?
There’s a version of this in in the DCMA takedown notice, which as I understand it is easier to file than to prove, and ends up with lots of false positives.
@WiseWyzard@KathrynTewson Yeah, I see what you mean.
In the us (major states I guess) can you file a lawsuit as an individual and what would happen if it were filed and not followed up on?
Someone trying to tip the balance of power could flood identifiable individuals with lawsuits.
@jontobinla What’s your goal, understand how to prompt and use the existing tools? Better understand the LLMs themselves? There are a lot of layers (ha, deep learning pun) you can dig into.
@WiseWyzard@KathrynTewson Yeah, I could see that as being challenging. That's well before the lawsuit stage. I'd be curious what @WiseWyzard was referring to.
@FounderKyle Did you start as a solo founder?
How did you find the solo vs co-founder experience different?
YC pushes co-founders hard, curious to hear of more on real-world experience.
@devahaz Note "feels" is an important modifier here. It can easily be argued that a lot of car crashes are systematic, but feel more accidental to most people.
i.e. Poorly designed roads/intersections.
@devahaz One thing that changes with a shift in error rates between "humans at 5*x%" and "tech at x%" is that human error is largely uncorrelated.
People have a bias towards uncorrelated risk. Car vs plane crash, one feels more "accidental" and the other "systematic".
@ryanrsamii AI has no keyboard, so it never has typing errors. But it does make conceptual errors.
For example, it will use the *wrong* function call (text_summarize instead of item_summarize), but it will never mis-type a function call (item_sumarize).
@ryanrsamii How deeply they review it will evolve over time as they get a feel for system's strengths/weaknesses.
For example, with copilot (AI for coding) I never review the code for *syntax*. There are basically never typos in it.
@LaserIsNotSound@abacaj Does GPT-4 decrease the likelihood of reform?
I bet it does. If everyone had to do their taxes manually on paper forms, we would already have seen reform. But @FreeTaxUSA and such make it easy enough I don't start writing letters.
GPT-4 may make things even easier.
@VRMink@svantetobias On 1. Fair enough, though I actually view not being directly responsible for my high-value "coins" as a bonus. (extreme example, but this: https://t.co/Braf0e1zXw)
@VRMink@svantetobias I'd argue 2 is mostly a "I'm probably doing illegal things". Heck, my experience is EU has awesome electronic transfer systems with very very low costs. US is slowly getting better.