> That's the mainframes, power systems, research, and what else?

None of it is part of GTS, cringely doesn't even know how the company is currently organized.

It's in fact very common for dynamic content.

Cannot be in the header, since the headers are not compressed in HTTP.

That's certainly possible, just like it's possible he tolerates 700 kilobits/second on what he thinks is 21 megabits per second and hasn't pursued it as some acute problem with his service, or just gone elsewhere.

TFA says they are able to come close for a random sampling of users, for what they claim to be a maximum speed for some class of connection. When I see someone post about "21 meg or some shit" and claim to be tolerating 700 kilobits/second in the early evening, I assume they're mistaken or haven't behaved rationally in resolving it.

Sounds pretty likely by the ambiguous units in your post that your expectations are inflated by a factor of 8 because you're misunderstand the units of what's advertised vs. the units in what you're observing.

Do you really think there's some abundance of extraordinary programmers that are incapable of thought, communication, and working in an office because they didn't spend 36 hours "studying" Philosophy in college? Can your barber or cleaning lady think, communicate, or "handle" living in society? Is this argument the result of some superior general education curriculum during your BS?

Everything in the story makes sense as describing an architectural style and some architectures that follow the principles of that style. Seems like the gotcha in the grandparent post is some mis-used can criticism of "is it written in REST?!". Extra lame points for trying to leave "architectural style" out to play up the difference.

Maybe you're on an OS with a dataready or HTTP accept filter. Timeout applies to reading the entire first line of the request.

aka Richard Pryor Art

well done!