erikaww

Is it possible to send and download documents as non-pdfs? I’d love to avoid the hellish pain that is pdfs

There should be legislation to give control over zoning and development regulation to HUD. Building just a quarter of the 2.4 million needed homes will take a minimum of $120 billion. We need some way of forging bipartisan support for fixing zoning regulations. I believe interest rates are coming down as well. If we can act fast, I think this will be a solved problem in a decade

If the index or incremental static analysis object was designed well enough, I don't think you would need to lint every file, you would just need to look at files that consume that variable. Maybe you would look at every index?

I'm not sure how well this could scale across (600- 1000?) different lints though. I should look into static analysis a bit more.

I'm not sure if Eslint has this, but there could be cross-file lints (eg. unused variables). If some file changes, you may need to relint dependencies and dependent files. This could recursively trickle.

I'm not sure if Eslint does this either, but indices or some incremental static analysis sounds like it could help the linter minimize rechecks or use previous state.

nyc in particular: 650 miles of bike lanes and 6300 of roads (not including lane-miles). thats 10x before considering the several order difference between the cost of roads and bike lanes. imagine if they just received 5% of the funding cars do

edit: turns out that source was old. they have 1500 miles of bike lanes, but my point still stands.

I think there is promise with tons of optimization problems (machine learning, approximating differential equations). I hope these would be significantly faster in practice.

It would be cool if we had quantum accelerated game physics engines. Super realistic games that are fast.

Simulations could scale better, so we could make them a lot more useful.

If only this energy and all the anger was directed in finding actual perpetrators of assault and charging them for their crimes. They are everywhere.

The right loves to fixate on this, yet they aren't doing anything.

isnt that wrt scaling size? couldn't they make other improvements?

i'd be real interested if they can rebut with big multimodal improvements.

Yeah and GPT-4V is even younger (at least it was available later)

yeah a lot of local models fall short on that benchmark as well. I wonder what was different about GPT3.5/4's training/date that would lead to its great hellaswag perf

the very smartest may not always seek the highest paying jobs.

it is at least true in sweden that the highest earners are not necessarily the smartest: https://fortune.com/2023/02/17/smartest-not-paid-most-wages-...

(i think this was on the front page recently)

clause 2.1 is $8/million tokens for prompt or .008/1k tokens. interesting

certainly more things to throw at the wall! Excited to see the "accidental" progress

Great way to find things and get more learning content out there (directly or indirectly).

How do you handle migrations?

I know the joke of this is that they could have in-housed this.

But why do government agencies not always have groups focused on financial efficiency? Is it corruption or stupidity?

This is amazing! I was just trying to figure out how to do this with drag and drop yesterday.

This could be useful for finding which region of Japan has an individual that can remotely kill with only a description of the victim.

kid orchestration may end up becoming a bottleneck

I am not educated about particle physics or whatnot, but isnt the uncertainty principle momentum and position cannot be both estimated to arbitrary accuracy? so one could estimate only one of them to arbitrary accuracy and then 1000 particles increases the confidence of that position estimate?

I'm not educated on QC. But wouldn't Grover's algorithm be very useful? I believe it provides a sqrt(n) time search. For QC to be useful,it just needs to make common utilities very fast. I should look more into QC. It would be cool if it could speed all computation up. I understand there are algorithms that it could execute that a classical composer could not, but I wonder if it could eventually he like just faster hardware.

There is a bit of work within the last 5 years with creating more algorithms for temporal networks (think graphs that change over time like social networks, neural networks (with evolving architectures) or economic chain networks). These are essentially streaming algorithms over temporal edgelists. I worked on a research project that visualized these networks in real time. Very fun stuff.