compiler-guy

"Goldman Sachs doesn't know what they are talking about" in reference to a stock maneuver is pretty unlikely to be true.

Even beyond GS, some of these companies are among the most sophisticated financial institutions on the planet.

The article says 22 companies have commented, but more have commented since it was written, or maybe the original article missed a few. Those companies include the list below.

Alphabet

AppLovin

Applied Digital

Applied Optoelectronics

Arm Holdings

Caterpillar

Cerebras

Cisco Systems

CoreWeave

Goldman Sachs

Intuit

Marvell Technology

McDonald's

Microsoft

Moderna

Qualcomm

Rocket Lab

UnitedHealth Group

Verizon

Wells Fargo

I didn't suggest that they revert to the approximate 2020 headcount. The question in the GGP was, "How much of the Covid headcount growth is there to revert?"

And the answer is, "Calculate what would have happened without Covid, assuming previous growth patterns, and compare that to current headcount."

So they would revert to the 2020 headcount plus a guess as to non-covid influenced headcount. A reasonable guess there is "about the same amount on average as pre-Covid growth."

The right answer here is something like:

Usually not, but sometimes under circumstances enumerated by law. Like this case.

Things during dotcom crash looked at least this bleak. Easily. It felt like every company on earth (at least in tech) was crashing and firing people. Even high-flyers like Google absolutely stopped hiring.

If you were lucky, you got laid off with a severance. If you weren't, you showed up to work one day and the doors were locked with no notice.

It is much better than optics friendly, because it gives the affected people a chance to determine their own destiny.

Layoffs absolutely suck, and there is likely more pain on the way, but if the alternatives are voluntary buyouts + layoffs vs plain layoffs, the first is much better.

Definitely a precursor to layoffs, but they also know that not everyone will take the offer. So if you want to reduce your workforce by X%, you offer (X+Y)% the buyout. The more attractive the offer, the smaller Y needs to be.

Personnel firms have decent experience predicting the takeup of an offer, so they make predictions about how good the offer needs to be, and they try to get close, within expenses and other considerations.

If Z people take the buyout, then, after the dust settles, you layoff the remaining X-Z people.

So it is unlikely that their target reduction is 25%.

There is enough until the company hits the early 2020 approximate headcount, adjusted for business changes in that time. The exact adjustment is hard to know, but in general, if a company was growing at, say, 10% per year, prior to the covid-craziness, then one would expect something like that natural growth to continue. Very generally speaking.

For example, Google had been growing somewhat quickly for decades prior to 2020, but in 2021 and 2022 it grew by over sixty-thousand people, and the crazy growth actually started in 2020, so even more than that. Sixty thousand!

Undoing that takes a very long time--they should have been much smarter about it. You need four rounds of twelve-thousand person RIFs (whether layoffs or buyouts) to get back to the earlier growth trajectory. Each round generates several stories that hit HN.

Is this MS still dropping back to pre-Covid? I would bet at least some of it is.

In my experience, the people who subscribe to type 1 are afraid that a potential Mozart or Beethoven won't reach their full potential because of bad habits and bad instruction. And the people who subscribe to type 2 are afraid that some kid won't have the pleasure of making music because they can't even start making music for years under strict type 1 tutelage.

There are many more people who never make music because it wasn't fun when they tried than potential Mozarts and Beethovens.

Yes, Google's money comes from advertising. But the company that invented, among many other things, map reduce, bigtable, and spanner, seems pretty much a technology company to me.

Bigtable these days handles seven-billion requests per second over several exabytes of data. All developed in house.

The others are similarly impressive.

https://en.wikipedia.org/wiki/MapReduce https://en.wikipedia.org/wiki/Bigtable https://en.wikipedia.org/wiki/Spanner_(database)

Yes. Those are clones or clonish. Not originals. So saying "Autodesk is opensource" isn't correct.

A clone of autodesk may be open source however.

Decompiled and easily available if you know where to look? Sure. Seems likely.

But "Open Source" is a description of the licensing terms, and it ain't gonna be that. Apple will sue hosters into oblivion. It will only be available on pirate sites.

The parent comment is a tongue-in-cheek comment that emphasizes the negatives of a location because the commenter likes it so much.

The idea (and it's a semi joke) is that it will discourage more people from coming so it doesn't get spoiled.

A bit like telling people some tiny restaurant isn't all that good, because if too many people eat there, you will never get a table again.

It's a tongue-in-cheek statement to prevent more people moving there so crowds don't spoil it.

Seems to me like it was an explicit design decision that fixed some use cases and broke others, rather than a regression in the sense of "A bug we introduced but didn't catch." As such a regression test wouldn't have caught this.

Someone made a design decision that broke some use cases.

... and there ain't no bugs on me. There ain't no bugs on me.

Except Apple can't even sing that verse.

The article does not distinguish between these two. Observe how the article talks about the laws requiring better maintenance result in higher hoa dues [1].

Nor do real estate listings [2].

And indeed, the grandparent post talks about missed water bills and other bills that fall on the collective rather than fines about incorrect colors or trash cans being left out.

So while legally there are differences between the two, the term "hoa" is almost always used as a blanket term to cover both. As they were in this article and nearly all of the discussion in this thread.

1 ... the laws that followed it require older condominium buildings to pass a milestone structural inspection (section 553.899, Florida Statutes) and to fund structural reserves on the strength of a structural integrity reserve study (section 718.112(2)(g), ...

The cost is visible in the dues. Redfin reported in August 2024 that median monthly HOA fees in the three months to July 2024, against a year earlier, rose...

2. None of the listings here distinguish between hoa dues and condo association dues. https://www.zillow.com/homes/for_sale/273801_rid/ Nor do several other cities that I tried.

And when the building's (that you share with the deadbeats) roof leaks and there isn't the money to repair that shared resource, that's just tough luck?

An HOA isn't just about neighborhoods of freestanding houses, it's also about multi unit buildings.

It isn’t “just” a relabeled weed burner, it’s a relabeled weed burner... with a fancy plastic shell.

And that makes a big difference, right? Right?

The last token ring standard appeared in 2001 and there has been a huge amount of network-technology development since then. Tokenring is tiny and incredibly niche. It may as well be dead.

Ethernet got its latest in 2022 and another is in the pipeline.

The difference between the standard and the implementation is useful for implementors, but makes no difference to users.

From a user's perspective, if the implementation sucks, the standard sucks.

I skip past them to the actual search results which tend to be much better.

Mike Levine (a financial blogger for Bloomberg) says the industry is calling these sorts of transactions "reverse acqui-hires" to distinguish them from a normal (forward) acquihire where the acquirer buys the whole company even if all it wants is the people. He discusses some of the advantages and legalities here:

https://news.bloomberglaw.com/mergers-and-acquisitions/matt-...

This reflects pretty much every result of mine from Google these days. Not just navigation, but search too.

The only thing I can figure is that Google's consumer-facing free-to-use models are set to "reason as fast as possible with only the most basic plausibility checks." I have to correct them like this every single time.

It's a little bit like any benchmark, especially microbenchmarks. It is useful as far as it reflects the broader use cases, but the risk of overfitting to the specific task and not the broader ones is very high.

I included the phrase “not their normal home” explicitly to cover cases like the Bodleian.

Unless you are arguing that the rare museum exhibition is the normal case whereas books and recitations are the abnormal ones.

Neither poetry nor literature is displayed in a museum (or at least that's not their normal home). But no one argues that they aren't art.

> "Everybody just complains ... and ... we say it's taking energy away from GIMP."

It's almost like Hacker News is a giant group of people, and different subsets of people comment on different subsets of articles (aka, the Goomba Fallacy).

Joel Spolsky has a famous post about this:

=-=-=-

A lot of software developers are seduced by the old “80/20” rule. It seems to make a lot of sense: 80% of the people use 20% of the features. So you convince yourself that you only need to implement 20% of the features, and you can still sell 80% as many copies.

Unfortunately, it’s never the same 20%. Everybody uses a different set of features.

https://www.joelonsoftware.com/2001/03/23/strategy-letter-iv...

=-=-=-

Also, there are many image editors _already_ out there of varying levels of power and ease of use. Some open source, some commercial.

And yet Photoshop still rules.

One that is all in on AI, and wants to be sure its engineers are too.

That's funny. In a meeting with my manager recently, they specifically called me out for not spending enough money on tokens. It's not like I didn't use it, just apparently not enough, and apparently on too low a setting.

Since then I've had fable cranked up to 11 for even the most trivial of tasks.