The source for the font appears here [2] with an Intel copyright.
[1]: https://wiki.softhistory.org/wiki/PhoenixBIOS_4.0_Release_6....
[2]: https://github.com/tianocore/edk2-codereview/blob/0a5f23c06d...
Plus, making the characters shorter would make them also a bit less legible.
Probably.
A lot of software back then was rendering UIs with box drawing characters, doing complicated layout etc - it's probably much easier to retrofit existing software to output the exact same thing in a slightly higher resolution font than to make the display logic support different resolutions properly.
Also remember that SETUP as a whole was often a mess of different ROM programs from different manufacturers. Far too many thought it their own ROM's business to clear the display and set text mode before they splashed their copyright strings, prompts, self-test, and hardware auto-detection information up.
Yes, they could have done things properly. The way to detect an existing screen size was there from the VGA onwards. And they could have left the display uncleared. And the whole preserve-my-graphical-splash-screen thing eventually did happen.
They did not.
There was also small .com utils circulating that would change to a tiny font so you could have 40, 50 rows of text with tiny fonts, or maybe even more.
The energy star logo was also displayed in text modes - by using custom font glyphs!
I really enjoyed Intel's desktop boards. They weren't particularly flashy, and there were certainly "duds" in there (early MU440EX revs not handling Pentium II CPUs properly, the whole RAMBUS debacle) but in general Intel made a solid (and not at all flashy) board.
With Intel's manufacturing competency you could be assured every board would be consistent. If there was a defect (I'm looking at you, MTH in the 820 chipset) every board would consistently have the same defect.
Is there any info or research on which width/height ratio of a fixed width font is the best for readability?