simsla

Source? I'm in Europe, and it's on my phone.

The hard thing for me is finding which questions to ask. Most of my 1:1s are just "err, want me to give you an update on what's been going on?"

Recently bought new curtains, and I selected on size, colour and made sure they were blackout curtains.

For the life of me (literally) I wouldn't be able to name the fabric.

Not everyone [1].

There's a voting mechanism to decide if comments aren't relevant/contributing.

[1] https://xkcd.com/1053/

Do you know a good source to learn about the FIM?

(Postgraduate level stats/maths, mostly applied, tiny bit pure.)

I've found that people also respond to thoughtful comments on blogs.

It's a DB. So you use it for storing data.

And merkle trees are what give "blockchains" their tamper protection: each state has a hash which includes the previous state's hash. (Although it'd be a list instead of a tree in the BC case.)

What's the use? I imagine 95% of companies that pretend to need blockchains could use something like this instead.

Funny, that. In my psychology 101 class, I was thought that random and intermittent rewards work better for conditioning behaviour than consistent rewards.

So in a way, the inconsistency might've pulled more people in.

There's also cost.

My company gives me access to a few journals, but at home I have no such thing. $20 is ridiculous for a paper, given that (a) the authors rarely see anything of this money, and (b) you often need to skim 10 papers before you find the 1 that's relevant.

Luckily, many papers in my research domain (compsci/ML) are open access. 90% is either on arxiv or Google Scholar knows a pdf URL.

For the rest, scihub is a lifesaver.

I expect I'll get infected eventually. I'd prefer not to get infected at a point when the system is (about to be) overloaded.

Omicron can't be meaningfully stopped, but it can be rate limited.

Today was my last day.

I moved to the UK for a pretty awesome job. These last few years have led me to re-evaluate my priorities, and I'm moving back home to be closer to family and old friends.

Many of my international colleagues (especially European) have made similar decisions.

> and especially on ability to fund companies easily via labor

Not sure I'm following. How would that work, and how would crypto facilitate this?

In the EU industry PhDs are fairly common. In this case, the PhD student does R&D work for a company, and gets paid through an EU PhD grant.

For some companies, it's an indirect way to get funding, because it provides for cheaper highly skilled labour.

(Anecdote incoming) For example, many years ago at the end of my industry internship, the company offered to continue the project as a PhD, pursuant grant funding. When I declined, they offered a well paid full-time job to do the same.

Dang is a Vietnamese surname (I know more than one "Dang"), which is why I also assumed he was Asian.

Could still be due to search personalisation.

I get programming results even for non programming terms. I am pretty confident that's not the case for my mom.

It stands to reason that people's search results could be wholly uninformative if Google thinks the subject matter is unlike what you usually read.

The problem with this (very popular) argument is that you can't give a CS course to a baby and expect them to get at programming.

By the time we see our first line of code, most of us have seen a ridiculous amount of data. We've been trained in problem solving, logical reasoning, maths, natural language processing, ... Hell, we've been trained as pattern matchers since we've been born.

By my account, humans actually need a large amount of training data. It might be the knowledge federation and generalisation that we're good at, but I don't think we're a clear winner in data efficiency.

TL;DR: making industrial (e.g. manufacturing) robots easier to use, by improving sensing, planning, etc.

I suspect that Dr. Chelsea Finn's work in meta-learning (affiliated with Stanford and GBrain, when I saw it last year) might play a big part here, which is e.a. about generalisation of RL policies to out of domain tasks. (E.g. similar task, but slightly different tools, slightly different task, etc.)

Learning IRL (cameras and actuators) reinforcement learning policies is a huge time sink, so generalisation is a hugely important task. Related solutions can be found in simulation->real generalisation, also an active topic of research.

Dang, just wanted to say thank you for all you're doing for the HN discussion boards.

Hope this doesn't break the rules for constructive responses. :-)

I'd only mention it if I could be considered biased, or if it's relevant to the conversation.

Disclaimer: it would indeed be silly in this comment

Instead of the "puzzles" thing, which doesn't seem to resonate with a lot of people, I think you're getting at a certain inquisitiveness.

People who say "huh, that's weird, I wonder why?"

I do agree that a top-down approach is preferable, but in the meantime I still see self-organising and collective bargaining as a good thing.

It's not just about wages. I think Amazon has actually made great strides by raising their minimum wages to $15/h. According to US stats the monetary benefit from unionising is about +11% on average. By offering twice the federal minimum, it's possible that Amazon is already above that. Now, you can argue that minimum wage should be higher still, but that's a separate discussion.

I don't think we should pretend that FC jobs will ever be super high paying or enviable jobs, but there are other ways to improve worker conditions. e.g. In some FC's, people allegedly have to walk for 20 minutes to reach a toilet. That's easily fixed. Making sure people don't have to choose between feeding their families and casting a vote in the election is another example. In general, even for a low skill job, pressure shouldn't be so high as to burn people out.

I think some FCs might actually have decent worker conditions. Others don't. As you alluded to, a company has financial interests that may not align with worker interests, which is why I think this needs to be approached in a structural way. Like Bezos himself once said: "Good intentions don't work; you need good mechanisms to make anything happen."

I hope enough people vote yes.

The way FC workers are treated is not ok. It's the main blemish on a company I otherwise respect. "Other companies treat FC staff like shit too" is not an excuse.

Disclaimer: Work for Amazon (corp / tech side). Opinion my own, obviously.

I really liked mother of learning. If anyone knows anything similar, I'd really like to hear it.

My point is that if Amazon e.g. directly integrated with any one of those solutions you'd enter an adversarial scenario, and all of the scammers would try (and almost definitely succeed, even if just by brute force chasing false negatives) in finding a way around it.

The same would happen if e.g. half of Amazon customers would use one of these solutions.

The difference between being online versus being in an adversarial scenario largely depends on how / to what degree you're visibly impacting the bottom line of the scammers.

I suspect that most of these problems are being actively worked on.

Unfortunately, while it's relatively straightforward to e.g. detect fake reviews in a static dataset, it's a constant arms race when you're doing so in an online adversarial setup. Same goes for counterfeit 3P goods.

Disclaimer: I work for Amazon, but not anywhere near the retail side.

Interesting. When I take Concerta (54mg) it makes me unable to sleep for the next 12 hours.

It doesn't make me any more awake/less tired than I normally would be, and I can lie down and kind of zone out, but I never completely drift off. Pseudo-meditation is the closest I can get to sleeping when I'm on Concerta.

It was a fake zoom link. Zoom wasn't involved in any stage of this hack.

Most of this is a discussion about semantics, which is silly.

In 99% of industry, AI is used as a synonym for Machine Learning. The remaining 1% refers to Artificial General Intelligence, which is a long term vision for what might be achieved using, again, Machine Learning.

Point still stands. UK software engineer median wages are above UK national median wages.