neilc

Does the task description influence the blocking behavior? That wasn’t clear to me— it might be that you manually configure the allow/block list and the task description is just for the user.

I don’t remember Amazon claiming “near-full automation” by AI. They said that you can checkout automatically and that AI/computer vision is somehow involved.

Maybe but that is also not within Apple's control.

They absolutely understood the terms of the loan. I think they do it anyway because: (1) everyone else is doing it and they don’t have a good alternative (2) career optimism. If you end up becoming a lawyer/consultant/accountant or otherwise have a good corporate job, in the medium-term your student loan debt doesn’t really have a significant impact on your life.

It’s really only a problem if you (1) choose a private college and don’t stay in-state, (2) get a degree which doesn’t have a lot of practical value, and (3) then want to pursue a low-paying field or get a not-useful graduate degree. For example, a friend of mine did her undergrad in art history, master’s in museum studies, and works for a non-profit. She’s not rich but she’s able to survive reasonably comfortably. She’s not dumb or financially illiterate, and she knew what she was getting in for.

I suppose we’ll have to agree to disagree. I know lots of liberal arts majors who still have a lot of student loan debt in their late 20s and 30s. They knew what they were doing when they enrolled in college and chose their major, it wasn’t like cost of tuition or what an “interest rate” is was somehow obscured from them or too difficult for them to comprehend. In some cases they regret the choices they made earlier but that’s a different matter, those choices were not made in ignorance of the basic situation they were entering into.

Basically all college-bound 18-year olds understand what debt is — you’re infantilizing them to a ridiculous degree if you think otherwise. A lot of them choose to proceed with college due to career optimism and following the herd, not because “debt” is some magical concept that they don’t understand.

It wasn't "good" when Google hired a ton of people during Covid, and it isn't "evil" when they have subsequently let some people go. The people who are being laid off are generally given very generous severance packages, and I think it would be hard to argue that Google treats its employees poorly in general.

Google should employ a workforce that they think meets their needs as a business, and when that involves letting some people go, they should do their best to treat those people fairly, which AFAIK they generally do.

Google makes a very small percentage of their revenue from government contracts.

Cycling is good. I've also found boswellia and curcumin supplements to be surprisingly effective for the arthritis symptoms.

It is definitely a thing that happens: https://www.nbcnews.com/tech/security/former-twitter-employe...

Hard to say how often / what fraction of occurrences are never caught, or caught but never publicized.

Also relevant are the difference between overall comp for software engineering jobs in the USA vs Europe.

Looking through the documentation, the API looks brittle.

Thanks for the feedback! Can you elaborate on the parts of the API you felt were brittle?

I’m pretty sure that people like John Carmack, Donald Knuth, or Fabrice Bellard are not spending money on PR firms.

You also need to factor in the time value of money: if the winemaker sells you a 2022 release in 2022, they get paid immediately.

Also factor in temperature and humidity controlled storage (a kitchen fridge will not do), insurance against disasters, backup power generation, and so on. If you think aged wines are overpriced, it is easy to cut out the middleman and age it yourself — so my guess is that the market is reasonably efficient.

Alameda is helmed by quants from Jane Street etc.,

The CEO worked at Jane Street for less than 18 months and appears to have had a fairly junior role there. I'm sure they are smart folks but there's a limit to how much you can learn in 18 months, in your first job after college.

Also not equivalent in terms of tax treatment.

The assets of the company are exactly the same, except that $1*num-shares that was previously on the company's balance sheet is no longer there. The company is less valuable and so the stock goes down.

Having an attractive dividend policy can make a stock more valuable to certain investors, but the act of actually paying out a scheduled dividend basically only makes the stock price go down.

Is the right to life no longer a human right?

That's not what "the right to life" means. There are lots of policy decisions which have tradeoffs that result in more or less life lost. For example, the government could require that all car engines have a maximum speed of 25 MPH. That would empirically reduce the # of lives lost in automobile accidents, but society has judged the tradeoff (in terms of convenience, transportation time/cost, etc.) to not be worth it -- and that tradeoff does not constitute "violating the right to life".

I’d love to hear more about your experience working in a professional kitchen.

I think the wireless pairing/device switching experience with AirPods is a big reason (not the only reason) why they are so popular.

Curious if you have any evidence for that claim. Looking at https://www.card.iastate.edu/ag_policy_review/article/?a=107 (Figure 1) suggests that rural vs urban unemployment rates have historically been quite similar; if anything, unemployment rates in rural areas were slightly higher than in urban areas. That changed during COVID but that appears to be a historical anomaly.

I don't think Bezos' ownership of the Washington Post has gone badly so far.

If your TAM and desired market capture implies revenue of less than like $500 million the company is never going to IPO, just for example.

The vast majority of tech companies that IPO have revenues of (much) less than $500M.

Furthermore, if you renounce your citizenship, it is you who are choosing not to get this benefit. It is a defined benefit for citizens

Individuals who renounce their US citizenship (or were never citizens) are still eligible to receive SS benefits; SS actually has very little to do with citizenship.

You do not need to be a US citizen to be eligible to collect US Social Security benefits.

https://www.aarp.org/retirement/social-security/questions-an...

Sure it does -- Apple's effective worldwide tax rate is ~25%. Like every US multinational, Apple structure themselves to optimize their tax burden, but it's not like they "have a history of not paying taxes".

Apple's effective worldwide tax rate is ~25%, which is actually slightly higher than other US multinationals. That is also not counting taxes on income/capital gains paid by California employees, which are obviously pretty sizable.

You might have missed that the benchmark was missing Nvidia (or even AMD) graphics cards;

The post does include a benchmark for an AMD GPU (Radeon Pro Vega II Duo) on the Mac Pro. Comparing the Mac Pro GPU vs. MBP M1 results, the GPU clearly wins, although in some cases the margin isn't as large as you might expect.

Hi William -- we have absolutely not copied any of Lightning's APIs.

In fact, our PyTorch API makes some significantly different design choices than Lightning does -- e.g., we require users to step optimizers and run the backward pass explicitly, which is a bit lower-level but allows for more flexibility when using the API.

For instance, here is an example of a GAN using our PyTorch API: https://github.com/determined-ai/determined/blob/master/exam...

This is a port of this PyTorch Lightning example: https://github.com/PyTorchLightning/pytorch-lightning/blob/m...

Despite the former being a port of the latter, there are significant differences between the two APIs.

More broadly, we welcome competition in this space and think there's a lot that we can all learn from one another.

Congratulations to the Grid team on the fundraise and the announcement! Exciting stuff.

It seems like there is an emerging consensus that (a) DL development requires access to massive compute, but (b) if you’re only using off-the-shelf PyTorch or TensorFlow, moving your model from your personal development environment to a cluster or cloud setting is too difficult — it is easy to spend most of your time managing infrastructure rather than developing models. At Determined AI, we’ve spent the last few years building an open source DL training platform that tries to make that process a lot simpler (https://github.com/determined-ai/determined), but I think it's fair to say that this is still very much an open space and an important problem. Curious to take a look at Grid AI and see how it compares to other tools in the space -- some other alternatives include Kubeflow, Polyaxon, and Spell AI.