I'm reminded of how IBM used to try to quantify coding for the purposes of promotion and bonuses:
It started with KLOC, which incentivized developers to write long, unwieldy code to juice up their LOC count.
So then it became Code Density, which incentivized developers to write a single line of inscrutably dense but functional code.
Finally they just gave up!
Yeah, so line count is not a measure of quality by any means whatsoever!
cc @pluralistic
Bill Gates called IBM's method of paying programmers by the line of code, "The race to build the world's heaviest airplane."
I believe the quote is from the mid-1980s.
@pluralistic @afx @evdelen @lizardbill
Other anecdotes:
1) IBM managers realized that programmers were just sitting there typing on keyboards, and getting paid a lot more than typists, for ( what they thought ) was a similar, slightly more technical job. So IBM starts firing programmers and replacing them with typists, with some training thrown in.
2) Airconditioning failed at an IBM office and programmers had to seek management approval to take off the ties ( they all wore business suits ).
@clayfoot @pluralistic @afx @evdelen @lizardbill
Ironic, since the first computer programmer was a woman ( Ada Lovelace ).
@purrperl @pluralistic @afx @evdelen @lizardbill
Huge culture change at #IBM was when they started wearing Polo shorts...
... Everyone was wearing Polo shirts so it became the new uniform
🙄
@n_dimension @purrperl @pluralistic @afx @evdelen @lizardbill
I was there when the memo came out that the board of directors had a meeting and they were formally changing the corp dress code to be no dress code. It took a meeting and a vote to decide to not so something. I had stopped wearing a tie months early when I noticed that half the folks in a meeting had ties and half didnt.
@evdelen @lizardbill @pluralistic
wasn't the K'Loc a Klingon battlecruiser?
@evdelen @lizardbill @pluralistic I'll never forget my 1st experience with KLOCs.
Boss: How many lines of code will this be?
Me: Um, Idunno, 1000.
later...
Boss: How many lines of code have you written?
Me: 250.
Boss: So, you're 1/4 done?
Me: Sure?
While KLOCs are completely useless, like every other software measurement, they had the advantage of being *quick* and relatively painless to produce.
@WildEyedBoyFromFreecloud @lizardbill @pluralistic
Manager: Why does this application load a Lorem Ipsum generator?
Coder: Line Count!
@WildEyedBoyFromFreecloud @evdelen @lizardbill @pluralistic : the point is not to produce *good code*. It never was.
The point is to built something shiny enough to be sold to Meta/Google/Microsoft because execs at those companies *think* it could become a competitor.
And for those executives, a product with a gazillion LOC is seen as "potentially scary so we should buy it".
That’s the whole business of Paul Graham. It is no software development. It never was.
You break the project down into functions, you eyeball the purpose of each function and guess how hard it's going to be based on experience, you write the functions. After the breaking down and estimating work, you have a good idea of how far you are through by which functions you've finished.
@Dubikan @lizardbill @pluralistic
@ploum made a great point here:
https://mamot.fr/@ploum/115021960673658871
TL;DR: Yes, arguably a Founder does have an incentive to produce a high LOC count.
@[email protected] @[email protected] @[email protected] @pluralistic : the point is not to produce *good code*. It never was. The point is to built something shiny enough to be sold to Meta/Google/Microsoft because execs at those companies *think* it could become a competitor. And for those executives, a product with a gazillion LOC is seen as "potentially scary so we should buy it". That’s the whole business of Paul Graham. It is no software development. It never was.
@evdelen @lizardbill @pluralistic
that sorta tracks with my experience. KLOC measurement was not as formal as this sounds but it was a thing. When I was asked about this metric I insisted on including comments. I wanted the folks writing for and with me to feel that well commented code had value.
There was also a number running around like the average programmer wrote 3 lines of code a day. Which was sorta true, we spent a lot of time in meetings. Design meetings good, mgnt meetings bad.