colonelxc

https://research.google/pubs/federated-learning-for-mobile-k...

I have a similar weather display made from the m5paper, but it doesn't look so nice (no graphics). It only updates once an hour though, and lasts over a month on a charge.

I agree, it is a great pattern if you can spare the memory.

From the website (https://s3fifo.com/), it claims that it needs no locking (backing scalability claims). This seems like an important part of their work too, unless I've missed some obvious trick that everyone uses. Naively, I would think that you can't update a hash table (to find the cache items efficiently?) and the queues at the same time without a lock. They surely aren't doing a linear search through the queue looking for a match

Thanks for mentioning this. I had looked at these before, but was holing off due to the cloud-polling home assistant integration.

I actually had a similar problem trying to park in a parking garage. I talked to some people working there (they were moving 'event day price' signs around). I asked if there was a kiosk or any other way to pay (other than by phone). They said there was not and said I just had to leave and find street parking.

This happened when I was trying to park near a place to get my phone fixed!

I've actually had that happen on Android, seemed to somehow kill even all non-tailscale traffic too

Didn't he also work on webkit at Apple?

There were record high # of businesses being created during the pandemic. There's anecdotes that it is because they had the opportunity to do so now with some extra cash + time off.

https://www.nytimes.com/2021/08/19/business/startup-business... https://www.npr.org/2022/01/12/1072057249/new-business-appli...

I had also seen some articles (citation needed!) that part of the employee crunch low-wage businesses are feeling is due to employees in those jobs taking their covid-money+time and training/interviewing/applying for office jobs.

I don't think we have a measure on how many people got to pursue their hobbies (rather than 'farting around'), though I'm not sure it matters. UBI proponents don't think that all people will do something 'valuable' though, just that it will be a net positive.

https://layoffs.fyi/

Interns are part of the recruiting pipeline and ramp-up pipeline. It gives the employer an evaluation period much better than an interview. It gives you a chance to see what work there is really like. They hope you enjoy it, and come back partially ramped up.

Speaking of ramp up, it's tough. It takes months for a new hire straight from college to be a net positive (not just when they start accomplishing things, but when they have accomplished enough on their own that it has 'paid back' the time spent teaching/mentoring/leading them).

Where does that put interns? Well, it's similar, but then the intern leaves when they're getting the hang of things (so generally they leave a deep net negative) Not to mention that interns generally are less experienced than the new hire, there's not expected to be a good return on investment during the summer.

Finally, because the intern has a lot of learning to do, and because they will leave soon, they basically can't be given any mission critical projects. So they will be given the low priority 'nice to have' tasks. This can result in a sad state where an intern will leave a 3/4 finished project, and come back a year later to see that no one has finished it (or worse, the project may be deleted).

(On the other hand, I think interns are a great way to get "low priority by PM standards" work done that the engineers really want to get done)

Economics? Well interns are cheap. Way cheaper than a FTE. They have low payment rates, and they're only paid for a few months. The bigger costs are the extra time from the engineers to help you ramp up (and the delays in their projects).

You'll need to just keep asking for tasks. Ideally you'll have a project that takes half or more of your internship. If you do, make sure to set milestones, break down the project into achievable tasks and demonstrate forward progress. At the end of summer you'll leave, and you want to have at least documented all the things you accomplished.

If they don't have stuff for you, see if you can take some tasks from their backlog. They probably wont let you have things that need to ship soon, but you can probably find something they'll agree to.

Good luck

They don't have to. The problem is that they said to investors repeatedly, "Sales are up due to increase in gaming demand", when they actually knew that a signficant portion of the increase was due to cryptomining.

So what the SEC is fining them for is lying to investors. If nvidia didn't know how much of their sales were to cryptomining, then there would be no fine. If they did know, and did share that in their reports, there would be no fine.

The "nobody dares giving bad feedback" thing isn't about retaliation (though I suppose that could happen). It's because perf is actually the worst place to provide "honest" feedback to a person about their performance.

It's complaining to managers/directors instead of talking to the person themselves (the recipient wont get to read your feedback for a couple months after). Even if you want to talk to a manager about some performance concerns, you should do that directly, instead of putting it in a record that sticks around for a persons whole employment

It's a bureaucracy game, and people who give bad feedback don't know how to play.

(I'm not endorsing the system at all, just rejecting the idea of it being retaliation-based. Anybody giving bad feedback doesn't understand what is going on)

In general, yes, I think that would be okay. I think it would be a mistake to create separate sections for each level. Overall your achievements within a single company should not be organized chronologically, but by what you want to show off. This may still be mostly chronological as you take on more responsibility/leadership.

Now, I don't object to adding a line like: "promoted twice from Software Engineer 2 to Staff Software Engineer" or whatever, which I think is a good middle ground (and I would put this as the very last, least important entry for that company)

I don't think that is unusual to list the latest level on a resume. I'm certainly not going to dedicate space on a resume to list time ranges for every promotion.

It's generally not required to 'permanently delete instantly' (this includes GDPR).

So you take a backup every day, and delete backups after 30 days (or whatever amount of time you need to for compliance). When somebody says 'delete this permanently', you do, from your primary data store. Then after 30 days, not even the backups will have anything.

You can also get fancy with more than one stage (soft delete, hard delete, then backup purge), as long as the whole pipeline is done within the compliance window.

Of course, the trickiest part of all of this is knowing where all data can proliferate, and making sure it doesn't (don't let people download data to their machines, don't let people create a 'copy of prod' that sticks around outside or retention.

It's still relevant. There are still numbers lower than 1%. Less than 1% of viewers will upvote/downvote a video they watched, which is much easier (and less expensive) than paying.

This might just be an impreciseness of language, but it's actually 20 days after you've been there 3 years, and 25 after you've been there for 5.

So if you're saying 'year six' starts the day after you've been there for 5 full years, then yes, we are saying the same thing.

(Source: have been at Google almost 8 years, and went through both of those bumps)

yes, though you should make sure you are allowed to install such software on your office laptop.

It has a tooltip on the loss section that mentions that the young trees need to get big enough to 'achieve canopy closure'

This study is about people who got Covid and now have issues, not issues with vaccines.

Most recalls are just to fix things. It might be replacing brake pads, or a bumper, or in this case some software.

I think the point is that the service mesh can't do the work of propagation. It needs the client to grab the input header, and attach it to any outbound requests. From the perspective of the service mesh, the service is handling X requests, and Y requests are being sent outbound. It doesn't know how each outbound request maps to an input.

So now all of the sudden we do need a client library for each service in order to make sure the header is being propagated correctly.

Advertising clearly wont "go die in a fire", no matter how frictionless the payments are. There was a conversation on hacker news just a few days ago about 'smart' TVs all getting ads, spying on the user, etc.

This is an example where people do have a way to pay for TVs (no need for microtransactions, TVs cost hundreds of dollars already!). But the TV makers have decided they can make more money by adding Ads, so why would they not?

This happened with cable TV too. You pay for the TV already, as a subscription even, why do you also get Ads? Because the cable company gets more money.

Can companies live while just charging for their services? absolutely. Will a lot of companies try to add additional revenue flows anyways? Also yes. In theory a company could compete on a 'no ads' platform. In practice, industry after industry realizes that they can just make more money at the turn of a switch. DeFi doesn't fix that. The advertisers are still going to come calling with their checkbooks.

I grant that DeFi does have some potential for micropayments that are hard with traditional finance. That could help make some blogs and small things ad free. But my point is that making payments has not at all stopped ads from invading every other industry. TV ads are not because your purchase had too much finance overhead. The advertisers will still be there, checkbook in hand.

A lot of people at the national labs have Q clearances, even if they never do nuclear stuff. As the article somewhat mentions, it's roughly 'Q=TS+nukes'. So if you need to do any TS stuff at a national lab, you're going to get a Q clearance.

From the footnotes:

"""Stablecoins that are purportedly convertible for an underlying fiat currency are distinct from a smaller subset of stablecoin arrangements that use other means to attempt to stabilize the price of the instrument (sometimes referred to as “synthetic” or “algorithmic” stablecoins) or are convertible for other assets. Because of their more widespread adoption, this discussion focuses on stablecoins that are convertible for fiat currency."""

I think you meant https://blog.cloudflare.com/branch-predictor/

There are a series of jokes in Google about things like this. Things like just wanting to serve a file, or run some small webapp, etc, etc. Google's tooling is geared towards building large, reliable apps. That makes wanting to build some tiny thing seem like a big bother (seemingly lots of overhead).

I remember my first hackathon at Google, probably in 2015. My group built a service and deployed it with some automated deployment tools. In "production", it was running 9 instances, spread over 3 continents. A little overkill for a hackathon project.

But I have never heard it (in 7 years) seriously.

For some reason XSS is missing from 2017 (should be 7), and as others noted, some categories merge (XSS is now part of injection)

As a googler, this is pretty embarrassing