Decision-making ability, judgment, and character can change a lot between teenage years and mid-twenties. Teenager's brains are not fully developed, and tend to lack the sort of reasoned judgment we'd expect from an actual adult.

Except that not everyone has done something embarrassing. Some children get lucky, others don't get into that sort of trouble to begin with. If we select for people who never got into trouble as children, we're losing a lot of potential.

Ideally, we try not to let teenagers mess up their lives permanently for stuff they do as children. It's not possible to succeed completely, but telling a 35-year-old that they should suffer from stupid posts they made when they were 15 does absolutely no good. Consequences need to be fairly fast to be useful in molding behavior.

What I want is the kids not being removed from their family twice for no good reason (like allowing stuff that was perfectly standard when I was a kid). That's going to traumatize the kids. Kids are resilient, but it's best not to disrupt their lives like that arbitrarily.

There's also a lot of hypocrisy. Quite a few people in government use illegal drugs and don't advocate for their legalization.

For developers interested in advancing in their companies, this also works the other way around. A good programmer with a decent basic understanding of the business and how it works is a lot more valuable than a good programmer who has to be told all the details.

The FDA isn't qualified to run the experiments it demands either. The FDA can look at the software practices and testing regimen. The other part is that medical software tends to have more stringent liability than other stuff, so there's a lot bigger legal risk for bad software.

One of the problems the US Air Force had in the Vietnam war was that North Vietnam's bridges were designed and constructed by people who knew they were bad at it, and so they were way overengineered and extremely hard to take out with bombs.

The problem with that is that the user has to go get something that's used by a relatively small number of people. In the early Hypercard days, all the stuff you needed to write programs was installed on your Macintosh, and there were lots of books about how to make Hypercard stacks. It was useful for some fairly good things, like the first version of Myst. I wasn't following it very well, but later Apple took the ability to write stacks out of the standard OS, then the ability to run them.

I used to work with a thoroughly mediocre programmer. She compensated for that by hard work and knowing her limitations, and accomplished some decent stuff.

If they build working, useful, and non-trivial SQL statements, I'd bet they could become decent programmers if they wanted to.

So what do you think of the AMA and various bar associations? I'd consider doctors and lawyers to be professionals, and they seem to like their unions (even if they don't call them that).

It's been going on for a long time. COBOL was intended to remove the need for business programmers, by making things so simple and self-explanatory that even a manager could understand them. Assembly language was called "automatic programming" back then, since all you had to do was write this text file and you didn't even have to assign your own memory locations.

Mythical Man-Month refers to the design and building of cathedrals, but it isn't the same sort of reference as ESR's.

The problem with Bitcoin is that it's an expensive transaction. Somebody has to do the proof-of-work to verify the transaction, and the proof-of-work is deliberately expensive. The only way to do it is to batch transactions, and that runs into the same problems as doing it with dollars or euros.