If the best available evidence indicates X, but you believe Y based on gut feel, then later solid evidence of Y is developed, were you right? Further, should this experience convince you to trust your gut over the best available evidence in the future?

If you read the article carefully, they are talking about lenses THINNER than a hair. I see several of the posts here thinking the width/radius of the lenses is this small, a reasonable mistake given the way this was written. Having a radius that small would severely reduce their light gathering ability, requiring very bright light or very dim images or very long exposure times.

-

Yep. Google, at least, started this transition during Trump1.

The company has long had engineering sites in various other countries, but until Trump1, the primary focus was always on cities where Google thought the global talent would want to live. Low cost was clearly not the driving factor in the selection of London, Zurich, Munich, Tokyo and Sydney, to name a few of the ones I visited. US sites were similarly not located in low-rent areas. The workforce was definitely global, because Google wanted to hire the smartest people and while the US does have its share of brilliant minds, the US has only 4% of the world's population, so most teams -- even in the US -- ended up being minority American.

During COVID, Trump leveraged the health crisis to essentially halt H-1B approvals and renewals. This caused significant problems for Google. My own team lost a few people because they couldn't get their visas renewed and had to go back home. Some chose to move to other Google sites overseas where Google could get them a work visa, others simply went back to their home countries. One trans woman on my team was in a particularly tough spot because her home country (India) refused to renew her passport because it didn't recognize her new gender. She couldn't get her visa renewed, couldn't go home to India, and also couldn't move to any other country with an expired passport. Luckily, she had a lot of nVidia and Google stock she'd been saving up to buy a house, and by cashing that out had enough free cash to get an EB-5 "investor" visa. It's good to be rich, of course.

Anyway, Google saw what was going on and, anticipating future troubles of the sort, refocused its overseas office plans on building up teams and infrastructure, especially in India which provided so much of Google's engineering talent anyway, with the intention of shifting whole projects and workstreams there. The company had long required a significant percentage of all staffing growth to be in the US (and especially in the bay area), but that policy was scrapped and replaced by its opposite: A certain percentage of all new roles must be based overseas.

It's still the case that the center-of-mass of Google is in the bay area, but the company is actively working to change that, to build up overseas capacity, and not just groups of junior engineers under a manager whose role is to pass them detailed requirements for implementation, but instead full teams with highly-skilled and experienced senior engineers and managers able to take full ownership of major product areas and move them forward.

Trump's latest moves will just accelerate this transition. The result will eventually be a hollowing out of the company's US capacity, and therefore a reduction in the need to hire American engineers. Lucky for me, I'm leaving Google for a startup and anyway am not far from retirement. Between this stuff and AI being poised to replace junior engineering staff it's a good time to be getting out.

Also, I think it will soon be time to start shifting investments out of the US.

It holds up in some cases, not in others, and calculating an average muddles that.

Personally, I use AI coding assists for two purposes quite successfully: a) more intelligent auto-complete and b) writing a piece of code using a common, well understood algorithm (i.e. lots of sources the AI could learn from) in the specific programming language or setup that I need.

It turns out that it is much faster and almost as reliable to have the AI do that then finding a few examples on github and stackoverflow, checking which ones are actually decent, and translating them myself.

Anything more complex than that and it starts being a coin toss. Sometimes it works, sometimes it's a waste of time. So I've stopped doing that because coding it myself is faster and the result better than babysitting an AI.

And when you need to optimize for a specific parameter - speed, memory, etc. - you can just about forget AI.

Hey, industry, I've got an idea: If you need specific, recent, skills (especially in the framework-of-the-month class), how about you train people in them?

That used to be the norm. Companies would hire apprentices, train them in the exact skills needed, then at the end hire them as proper employees. These days, though, the training part is outsourced to the education system. And that's just dumb in so many ways.

Universities should not train the flavour of the moment. Because by the time people graduate, that may have already shifted elsewhere. Universities train the basics and the thinking needed to grow into nearby fields. Yes, thinking is a skill that can be trained.

Case in point: When I was in university, there was one short course on cybersecurity. And yet that's been my profession for over two decades now. There were zero courses on AI. And yet there are whitepapers on AI with me as a co-author. And of the seven programming languages I learnt in university, I haven't used even one of them ever professionally and only one privately (C, of course. You can never go wrong learning C. If you have a university diploma in computer science and they didn't teach you C, demand your money back). Ok, if you count SQL as a programming language, it's eight and I did use that professionally a few times. But I consider none of them a waste of time. Ok, Haskell maybe. The actual skill acquired was "programming", not a particular language.

Should universities teach about AI? Yes, I think so. Should they teach how to prompt engineer for ChatGPT 4? Totally not. That'll be obsolete before they even graduate.

So if your company needs people who have a specific AI-related skill (like prompt engineering) and know a specific AI tool or model - find them or train them. Don't demand that other people train them for you.

FFS, we complain about freeloaders everywhere, but the industry has become a cesspool of freeloaders these days.

The number of people isn't the problem here.

The "started every" is.

How did they not catch that during development and found a solution? I mean, the meme's where a TV ad starts Alexa and orders 10 large pizzas are a decade old now.

Cppcheck doesn't flag anything in Waffle Iron's example.

It also doesn't find anything wrong with:

std::vector<int> vec = {1, 2, 3, 4, 5};
auto it = vec.begin();
vec.push_back(6);
std::cout << *it << std::endl;

Which is another common example of how you can write memory errors without using C++ pointers.

Sure, so C++ is safe as long as it's used in a way that makes it incredibly painful. Sounds good. Let's just require all C++ code everywhere to be written that way. Rust usage will skyrocket overnight.

He unintentionally used Signal? How drunk was he?

For a really rigorous definition of "fair", it's impossible to have fair democracy at all. Arrow's Theorem demonstrates this to a large degree, although many have argued that some of his fairness axioms are excessive. More recent research has concluded that fairness is the wrong standard, because there's no way for an electorate's "will" to really be fairly represented by any electoral system, not in all cases. Some systems can do better most of the time (and "winner take all" is particularly bad), but all systems fail in some cases.

What we need to aim for instead of fairness is "legitimacy", which is more about building broad acceptance of the system than about fixing the system itself, though it's easier to build acceptance for better-designed systems.

Having the country's top politicians continually claiming the system is unfair and rigged is, of course, the worst possible thing to do if you want to build support for the legitimacy of the system.

I decided to check :-)

According to the Book Censorship Database from the Every Library Institute, both "Of Mice and Men" and "Adventures of Huckleberry Finn" have been challenged, but only "Of Mice and Men" was removed, though "restricted" is more accurate. The Birdville Independent School District in Texas removed the book from general access, allowing access only to the AP English class, and the Indian River County Schools in Florida restricted it to high school students.

No Doctor Suess books were banned, although Suess Enterprises voluntarily ceased publication of six books.

Ah, my favorite among such is the warning from a hair drier I bought some years ago. It said: "do not use while sleeping."

Hmm. Lost the end of my sentence there somehow:

And, of course, the companies that are going out of business are also the ones who are least likely to spend time filing out a survey

The BLS monthly numbers are always off when the underlying economy is changing rapidly, because of the "birth death problem", meaning that when large numbers of companies are being created or closed (born or died), the surveys that provide the quick data are guaranteed to be quite far off because the surveys go to companies that are already establish, i.e. those that weren't just born and didn't just die. So when there's a lot of market change, they're sampling the part of the market that is changing less. This means the estimates are off, and the faster the economy is changing the further off they are.

A related issue is that the survey results are only a sample, but BLS needs to extrapolate to the entire population of businesses -- but they don't actually know how many businesses there are in the country, much less how many fit into each of the size / revenue / industry buckets. So their extrapolation necessarily involves some systematic guesswork. In normal, stable economic times good guesses are easy because it's not going to be that much different from the prior year and will likely have followed a consistent trend. But when the economy is changing rapidly, that's not true, so the guesses end up being further off the mark.

Second, it's worse when things are turning for the worse, because of something kind of like "survey fatigue", but not. The problem is that when lots of the surveyed companies are struggling, they're focused on fighting for their existence and don't have time to bother filling out voluntary government reporting forms. It's not that they're tired of surveys, but that they just don't have the time and energy to spare. And, of course, the companies that are going out of business are also the ones w

The phone thing is a red herring, because these BLS surveys are not conducted over the phone.

A new issue compounding the above is that the BLS was hit hard by DOGE cuts and early retirements. They've lost over 20% of their staff, and the loss in experience and institutional knowledge is far larger than that, because the people who were fired and the people who took the buyouts tended to be very senior. So a lot of the experience that would be used to improve the estimates has walked out the door.

Anyway, the core problem is that the economy is going into the toilet, really fast. The BLS didn't break out how much of the 911,000 fewer new jobs were added 2024 vs 2025, but I'll bet a big percentage were after Trump started bludgeoning American businesses with tariffs. Most of that pain won't really be known until the 12-month report next year, because the monthly reports are going to continue underestimating the rate of change. Well, assuming the BLS staff isn't forced to cook the books, in which case we'll just never know.

If I knew that, I'd be a tech billionaire inventing them.

The best part of innovations is that they are NOT obvious, but once they are in the world, you can't do without.