cf

I think we are agreeing. My point is you can get away with explaining Metropolis-Hastings in a useful way and never mention detailed balance. You will need it to work out a formal proof of convergence, but you don't need it to understand or even code up the algorithm.

I think explanations don't require formal proofs to be useful.

While I think detailed balance is typically how we explain why Metropolis Hastings works at all. It's actually not quite what matters. What matters is ergodicity and detailed balance is a proxy for checking for that. That's helpful since ergodicity does have an easier intuition behind it as well.

Many times things like comments are dropped and names are disambiguated. Printing the tree back out would yield an equivalent program but the source code would look way different.

My way or the highway doesn't work in politics. I'm sorry if you were told otherwise.

Alright, please see this as an olive branch to YIMBYs. The results are pretty clear. The YIMBYs do not have the votes on their own to get this done. Unless you are some suburban retiree you probably recognize that the solution is building more housing.

I think based on the voting records YIMBYs should join forces with PHIMBYs and make the next iteration of SB50 really take public and affordable housing seriously. The existing provisions added to SB50 already got the bill much closer. Why not add some provisions from AB1279? Maybe repeal article 34 in the constitution? Maybe repeal Costa-Hawkins? There are NIMBYs that are pretty clearly hiding behind affordable housing that we know are insincere. We know their cities and counties aren't doing enough to help poor folks. But the evidence suggests there are enough people that want to upzone and build housing to get this done.

Does this strike others here are a reasonable path forward?

This is old news but one thing I think Google misses. No one product they cancel is used by tons of people, but the amount of people fucked over by the cancellation of any one product is enough to be an ongoing reputation hit for them.

If you read HN you can figure it out. If you see the fine print on lobste.rs you can see there is a possibility to ask in their chat.

On some level it's not open invite and you have to be ok with that.

Just updated my profile with it!

Oh crap I didn't realize it wasn't public in my posts. Hold on! Also you don't seem to be an asshole! Invite sent!

If you want an invite, email me with your username or some other places where you've commented and if you don't seem like an asshole I'll hook you up.

I think using numpy is always good first step after just trying to improve the algorithm. Numpy will be less effort than going to cython. After that cython is a good next step. I seriously don't know any situation where I would do the kind of micro-optimizations mentioned in the article.

I think for pure numerics using numpy basically recovers all performance loss. It's when you get tricky control-flow that you run into real trouble.

That's basically Collabora

This is a pointer that's a bit buried. For all the fanciness of different particle filtering algorithms, you are likely not using enough particles. Unless you having troubles related to running out of memory consider using more particles. Even doing fancier things like rejuvenation and stratified sampling benefit from more particles.

No this has actually been measured by people.

https://danluu.com/input-lag/

The computers are faster, can do more stuff, and monitors have higher frame rates. But for many applications that aren't games latency and non-responsive UIs are a growing problem.

Intuitively for me probability theory is a bit clunky before measure theory. Like we have a different equation for expectations if something is continuous vs a discrete distribution. We use probability density functions (pdf) vs a probability mass function (pmf) and etc.

We know this stuff is basically getting at the same underlying quantities. Now imagine a distribution over both continuous and discrete. For example something that measures temperature but breaks after a certain threshold. What does the expectation be for such an instrument? Imagine a distribution on different sized arrays of real numbers. How do you define a valid density function? Measure theory gives you the formalisms for those kinds of problems. You in practice don't need it very often but it keeps you on firm ground when you do.

As you can see the number is going up. I would also be careful with the HUD data as it is based on a yearly count conducted of shelters in January [1]. Counting homeless people is tricky as only 10% are on the street and the methods used to estimate populations are fairly unsophisticated.

From the same surveys it's fairly clear in recent years homeless populations are rising if you look at exhibit 2.5 on page 25. [2]. Worse transitional housing is being replaced with shelter beds meaning those who are homeless are more likely to stay so.

I also should be clear looking big picture like this misses details. Things like Los Angeles's homeless population increasing by 16% [3] and a 5% rise in Seattle [4] and similarly Vancouver [5]. All of these cities have great GDP growth. Experts across the political spectrum are fairly consistent that the cause is related to affordable housing.

Now I can't make you change your mind about something you want to believe, but when you really engage with what's going on suggesting causality between GDP and homelessness is a very shallow superficial analysis. I think the data makes a pretty clear of what's going on. I'd encourage you no matter what position you choose to take to really engage with the details.

1. https://www.washingtonpost.com/business/2019/09/18/surprisin...

2. https://files.hudexchange.info/resources/documents/2018-AHAR...

3. https://www.latimes.com/local/lanow/la-me-ln-homeless-count-...

4. https://www.city-journal.org/seattle-homelessness

5. https://www.vancourier.com/news/vancouver-s-record-breaking-...

Homelessness has been increasing rapidly in recent years across North America even while GDP growth has been adequate. You actually need to devote resources to fight homelessness.

Well that just means you capable of having good ideas. Just keep at it and I'm sure you'll do something great.

I expect John Carmack to follow a trajectory of someone like David Ha who with little previous background started to write very creative and thought provoking papers (https://scholar.google.com/citations?user=J1j92GsxVUMC&hl=en)

It won't be AGI by most definitions but I bet it'll be pretty cool and I'm happy to have that.

From your remark you'd think these were two entirely different problems. I'll try to say it more clearly. There are people deliberately promoting fear of long-term problems as an excuse to not address these short-term problems.

I'm ok with people working on both. I just feel so much less attention is being placed on these more immediate problems. Tesla has deployed to production software for driving that is unable to detect pedestrians. So at least Tesla can't do both.

I think applying agency to the software in my examples is actually very pernicious. It's not the software that dislikes the ethnic minority, it's the Chinese government. Blaming the software lets bad actors off the hook.

Perhaps what's really needed is ethical standards for technology. This would have the added advantage of being applicable even if the technology doesn't feel like AI to some people.

The real risk is that worrying about fantastical disaster scenarios distracts us from addressing the more immediate problems we already face with AI. Whether it is facial recognition being used by China to aid in ethnic cleansing of a minority group, or Tesla Autopilot regularly killing its passengers.

Even if you don't like my particular examples I can highlight a dozen more problems AI software has created or exacerbated right now. Why all the focus on hypothetical problems?

The original awk book from the creators is in my opinion really good. https://archive.org/download/pdfy-MgN0H1joIoDVoIC7/The_AWK_P...

I make use of https://github.com/strukturag/spreed-webrtc . I self-host and this gives me the ability to offer a high-quality video or voice stream for somebody. While I still have some bugs when contacted from an iPhone, it has been a generally easy option.

Alternatively, there are webrtc solutions like talky.io and appear.in which work well enough. I have never had a client have an issue with these solutions, and which I choose is more a function of how much I trust my own setup with the goal being to move all interactions to my self-hosted solutions.

What I remember being really weird is those books were always next to all these books on psychics and UFOs.

There doesn't need to be a limitless supply. In many cities if buyers aren't found, property owners are content to leave them vacant. You can see this in the relatively higher rates of vacanies in condos compared to other kinds of housing.

Old units in many places are frequently renovated and re-sold to upper-middle class and wealthy tenants. The end result is an erosion of housing stock for average and low-income residents.

Thanks for taking the time to reply.

I am 100% in support of software like this being open-source. Not just because companies go out of business and disappear, but for scientific-computing software, it's important to be able to extend these systems. When I talk about features, I often mean can I rewrite this integral into something more computationally tractable? Or is there some reparameterization of this function into something differentiable?

I know this are very much how someone might pose a problem coming from Mathematica or Maple, but really I hope it's clear the end goal is still the same. Discover an equivalent mathematical representation of my problem that's more amendable for further work.

I know this isn't the most appropriate place to ask, but what are the main feature gaps between commercial computer algebra systems like Mathematica/Maple and systems like Axiom/Sympy?