RAM disks are cool and all, but except on live CDs they're usually unnecessary. The kernel's buffer cache and directory-name-lookup cache (in RAM) can often outperform RAM disks on second reads and writes.

(Claimer: I worked on file systems for HP-UX, and we measured this when we considered adding our internal experimental RAM FS to the production OS.)

At least #4 is a legacy from the ARPAnet and USENET from the same time period.

Nobody comes here any more. It's too crowded.

Remember, though, NYC is much more than Manhattan. In Brooklyn, Queens and the Bronx, you can be miles from the nearest subway station, and Staten Island isn't connected to the NYC subway system at all.

(Wow, I wonder how many hours have been lost finding bugs resulting from syntactically-correct typos like the one I just made...)

Column-independent syntax and free-form comments are a good start. (You can't enforce good commenting, but you can at least enable it.) And to be fair, COBOL's long variable names were a huge improvement over old versions of FORTRAN and BASIC.

COBOL initially claimed to be self-documenting because of its English-heavy syntax. Indeed,

    PERFORM DO-SOMETHING VARYING X FROM 0 TO 10 BY 5.

is much more readable at first glance than

    for (i=0; i=10; i+=5) { do_something(); }

but this let programmers think they didn't have to add many of their own comments. Thus, it was more likely that a typical ugly hack would not have been commented.

Teletype ribbons are pretty easy to replace, because plain typewriter ribbon can be loaded onto its pair of generic spools (no, I'm not showing my age here, nope). Try finding the ribbon *cartridges* for 70's/80's vintage dot-matrix printers. They were as device-specific as toner cartridges are today.

It's very easy to tell *what* a COBOL program is doing, but unless it's very well written, it's not at all obvious *why*.

Mmmm, EBCDIC...

20. It's been more years than that since I had that few. Since I'm not administering an entire development lab any more, I'm down from about 300 to about 100. That's about 100 passwords conforming to about 90 disjoint sets of length/alphabet/aging/reuse policies.

My dream is to have easy two-factor authentication into a vault full of strong keys.

Then why does bariatric surgery work?

Does this make Linus the Gordon Ramsay of software?