It was absolutely not the case that "there was always an eye towards how would this get deployed to a market as a product". Look at how much energy went into building Belle, the first chess machine to achieve master-level play. And it wasn't a simple computer program, it had a full-on chess board with pieces. When dignitaries toured the labs the tour guide would have to invent some bullshit reason as to how this chess machine benefited a telephone company.

Or just read the biographies and interviews of the people who worked there. They'd work on things they found interesting and there was no pressure to make commercially viable stuff. Now it sometimes happened that for budget reasons they'd work on particular projects, e.g., the computer science team got the AT&T patent department to pay for a PDP-11 with the promise that the computer science team would develop document preparation software (funnily enough, this same excuse was used elsewhere, e.g. at the University of Nottingham the computer science department got money to help with typesetting exams).

I don't even know what you mean by "long-term tech development is heavily done in an academic setting." When I look at papers from, say, SIGGRAPH these look like exactly the kinds of things you'd get from a handful of grad students and a PI, i.e., work that can be busted out by a few people in a couple years. When I think of "long-term tech development" I think of something like all the work Google did in AI which spanned decades. The projects that were done by Bell Labs in CS mostly took a few weeks and were done by one or two people -- like how much work really went into the first version of "pic." Most of the Unix work at Bell Labs was passed over to the commercial group (AT&T's Unix Support Group, as well as licensees which fed their work back to AT&T) after Version 7 was done.

Unless you do something stupid of galactic proportions like using the Macbook Neo to farm Chia tokens, the RAM and SSD will last the life of the computer. What will break are: battery wears out, keyboard and screen break, USB ports break. These are cheap and easy repairs. Apple is intending to disrupt Chromebooks from schools, so it is likely that district-wide there will be enough repairs needed to justify having an employee who does internal repairs. These are much easier for such an employee to do than having them all farmed out for Apple to do.

If that's the best excuse he can come up with, the PC laptop industry is fucking cooked. No it's fucking not a "content consumption" device -- most people use their laptops to use Chrome and MS Office or the free equivalent (which, incidentally the Macbook Neo comes with ... no PC laptop comes with MS Office anymore). They use laptops to do schoolwork, to send e-mails, to do their taxes, to shop online, all the stuff normal people have done with a computer for the last several decades.

The hint that it's not a "content consumption" device is that it ... wait for it ... has a keyboard attached. No it doesn't have the horsepower or RAM to do heavy duty software development or video editing, and the display is not quite good enough for professional photo editing, but those are a tiny minority of users, except among Youtube creators. Oh and yeah, it's bad at gaming. Funny thing though - the PC market for gaming is so unhealthy that Sony has announced that its titles generally won't get PC ports anymore because they sell so poorly, so that probably doesn't matter.

I really wonder who it was inside Apple who made the brilliant proposal that maybe instead of positioning its products as luxury goods (Angela Ahrendts was the worst hire ever) it should use its massive scale and efficiency to dominate the market at every level.

When a company is in decline you can either wait until the company is completely broke and it can't misses payroll (or misses payments on its loans), or it can hit the big red button and get bought out by a company like Broadcom or Bending Spoons that cuts costs and sucks the marrow from the bones. While it sucks for the company to die like that, it's better than the alternative and nobody should have been caught off-guard. Unless you were specifically paid a retention bonus to see the transaction through, your resume should have been updated the day you heard about the transaction.

Private equity is different because private equity intends to on-sell the company within five years (they get five year loans, and refinancing will be hard after PE has damaged the company). PE comes in to a fundamentally sound company, makes bone-headed changes like cutting product quality, skims a bunch of cash off the top and then sells the company. Completely different from a grim reaper company like Broadcom or Bending Spoons.

The worst part of Workday is that they transmit your personal information to pretty much every other company out there. Ever notice that when you go to Docusign it knows all the e-mail addresses of your company employees? Have you ever gotten calls on your personal cell phone from vendors you have absolutely not given the number to? Have you noticed that you start getting very specific spam IMMEDIATELY after joining a new company and you haven't even updated your linkedin yet? The privacy violations are out of control and even when I've talked to the head of privacy at companies where this is happening, they admit they don't have control over it.

As far as managing HR files I don't give a shit one way or another. There's got to be some software to manage HR data and Workday works as well as anything else. But the way that they misuse the personal data of US employees is something that regulators need to crack down on hard.

Big endian support is dead. Power was always great for testing code, because you could get an old Mac off e-bay for $100 and make sure stuff worked right. Now the only Power architecture supported is little endian.

I mean, if you see computer programming as just a means to an end, that's fine, but if you see it as an art form (c.f., "The Art of Computer Programming," written by some guy) it's important to write portable code.

We have a non-truncatable currency system. An ideal coin system will have the following characteristics - that it is "countable" for any number, and that eliminating the smallest coin will always leave the remaining coin system countable. "Countable" means that that cashier's algorithm of pulling the largest coin less than the amount needed and then repeating with the next largest coin will produce the optimum number of coins. For example, with our current system, if you need to make 42 cents, you do a quarter (leaving 17 cents), then a dime, (leaving 7 cents), then a nickel and then two pennies. Our current system is only truncatable for the penny. If you only consider coin amounts which are multiples of 5 (since other ones become impossible after getting rid of the penny), our system minus the penny is clearly countable.

The problem is, consider getting rid of the nickel. Now try to make 30 cents. You pull a quarter, leaving 5 cents, oh shit you made a mistake. Back up, you should have done three dimes.