Joshua Otti

I heard an engineer who described AL as a bunch of linear functions they don't understand. And I said no, it's a lot of very non linear stuff that nonetheless seems to act "linear" in some sense anyway. The non-linearity is actually very important.

Frankly, it should lose you your job.

It's a tough domain for tech. We are used to developing something alone, scaling it endlessly for free, and consumers trying it out at little to no cost.

Long term support and deep partnerships are hard for tech. It's a culture issue.

The computational model of quantum mechanics is straightforward, and classic computation and measurement effects arise naturally.

I don't know enough about the physics version of quantum mechanics to comment on that. But it feels like people are too obsessed with the classical world existing.

But the claim isn't even true. China is also investing heavily in AI and is trading blows with US companies regularly. They just aren't burning down their economy to do it.

As someone who just went through the hiring loop recently, CS hiring is crazy right now. Crazy bad, it doesn't make sense.

Who's up for building a second, much taller space needle at the other side of the park? Maybe where the memorial stadium is?

I think highways are good for first responders and emergency evacuations.

I'd start with congestion pricing.

It would also help me personally because my bus would make it on time due to less traffic. Quite likely it would also increase bus demand and thus frequency of busses in general.

It would also make me a bit less hesitant to drive through the city if I needed to.

I've said it before and I'll say it again: Seattle needs to implement congestion pricing.

It depends on the family the problems are drawn from. For constraint satisfaction problems, there is a dichotomy theorem:
en.m.wikipedia.org/wiki/Schaefe...
From these families, it's very likely one includes a too complex kind of constraint and we get NP completeness.

While it seems that way, it's actually the case that if P!=NP then there have to be NP intermediate problems. I think the proof constructs such a problem (although quite artificially).

That said, we have few candidate natural intermediate problems, such as factoring and graph isomorphism.

Most problems known to be in NP but not in P are NP complete. Few (natural) NP intermediate candidate problems exist that are well studied.

It is standard in complexity theory classes to prove that if P!=NP than NP intermediate problems exist because it seems like a dichotomy at first.

It's what investors want, and isn't that what really counts?

Three logicians enter a bar.
“Beer for everyone?” asks the barman.
“I don't know,” says the first logician.
“I don't know,” says the second.
“Yes!” answers the third.

*UI

Setting up VPCs is perhaps one of the most annoying parts of making any web service. I don't know why AWS can't make a better YI for it

Springer publishes a P ≠ NP "proof" and Eric Allender has words to say.

blog.computationalco...

Full time jobs really eat into your time to do research. Now I'm really facing the consequences of writing very long papers.

On the bright side, I finally wrote out the proof that Reed-Solomon codes cannot be encoded or decoded time-space efficiently.

Capital gains should be taxed the same as income.

I've been using Memrise for the past 3 weeks to learn igbo and it is way better than igbo drops. It teaches more phrases and multiple speakers per phrase. It could be better still, but it's the best igbo language app I've seen yet.

I think gentrification is only a problem if there is already a major wealth disparity between the people who live in an area and the people who own that area.

That is the root issue that should be addressed. Gentrification brings in a lot of money. It's a lever to fix the problem.

Cons:
1. The current people there might not afford it.

This is a separate issue. New growth can be new jobs, only an issue if people don't own property, etc.

2. It changes the 'character' of the city.

Some changes are good. If done wrong, there can be overreach. But cities grow or die.

The pros are clear:
1. Higher density, less commuter time, carbon emissions, etc.
2. Generally safer areas, higher quality of living for residents.
3. Due to higher density, less dense housing is not as far away, so even suburbs are closer and more affordable.

I'm not entirely sure gentrification is bad. To my understanding, it's basically replacing buildings in poor areas near city centers, with nicer, newer buildings with higher density while making the neighborhoods more walkable.

Which on the face of it seems good.

Yes, but I suspect that they will be much more efficient by then. I suspect that big big models will still be state of the art, but when properly tooled, small models will have close enough capabilities in what we care about that we use them, and they can be run locally.

What can I say, hints and fun facts are usually helpful for solving problems. LLMs use a very simple, dumb attention model. It's hard for them to NOT pay attention to the irrelevant slop.

It's also just a terrible excuse. You don't "know him well"?
1. You should know his policies: he's the dem nominy for mayor of the largest city in the state!
2. The average voter doesn't know him personally, and they have to decide. Why can't you?

I was using an igbo language app and was in a section called "basic necessities".

The first phrase was "do you have internet?"

A half full 2 literally cherry cola in a plastic device built to hold 8 2 liter bottles. Also has a plastic bag and a takeout container with it, all on a wide bus seat.

Things left on the seattle bus.