Oct 21, 2026, 8:14 PMRE: https://scoat.es/@sean/117480478803436020
I wanna say something a little more nuanced here:
It is easy and cheap to produce software that *mostly* works. It is also easy and cheap to learn enough programming to produce programs that are good enough for your own personal use.
The extra 90+% of costs that Mr. Coates is talking about here are *all* in the gap between "mostly works, most of the time, for me and people like me" and "works for sure, always, for everyone."
Oct 21, 2026, 8:19 PMFor the most part, programming isn't all that difficult. There are things that require heavy mathematical analysis to get right, but nowadays you can usually rely on someone having done that for you and wrapped it up in a library.
The catch, though: Writing a program that works for sure, always, for everyone, requires enormous amounts of fiddly, tedious detail work, and you need a lot of experience to be able to look at code, or program behavior, and see where that detail work hasn't been done.
Oct 21, 2026, 8:22 PMFiddly tedious detail work is no fun, and it's hard to persuade management it's worth doing, particularly if it means there's fewer developer hours available for shiny new features.
And this is one of the most important practical reasons why all software is bad.
Oct 21, 2026, 8:25 PM@zwol Where's that news story about someone who was like "UGH, why is my local government taking so long and talking about so much money, for putting five steps to make a path down a small hill?? I'm just gonna get some two-by-fours and bang together some steps myself! There! I did it for you!"
and then the follow-up reporting about the quality, safety, accessibility, usability, and longevity, issues that this made clear.
EDIT: one piece of the reporting, enough to give search terms for more in-depth from different angles (this story is a relatively soft touch, trying to vaguely express multiple people's perspective)
https://globalnews.ca/news/3614167/tom-riley-park-stairs-removed-toronto/
Oct 21, 2026, 8:29 PM@gaditb Which points at another important thing: programming ain't special. *Every* engineering discipline is like this.
The price of civilization is eternal maintenance.
Oct 21, 2026, 8:30 PM@gaditb (although it does make a difference that a CPU is one of the most complicated machines humans have ever constructed *and it doesn't do anything by itself*, you have to hook it up to multiple other things that *also* qualify as "one of the most complicated machines humans have ever constructed" to make it useful)
Oct 21, 2026, 8:42 PM@zwol The other difference is that programming as an engineering discipline has not yet come to face and account for the actual impact on people's lives we have, and take responsibility as an industry for failure and for accredidation.
(And like, other industries are at different points -- like, the difference between crash investigations in the airline industry and crash investigations in the railroad industry is a famous difference of proper responsibility -- but I think we're undeniably at the far back of the pack.)
(Even taking into account the "interviewing other-fields engineers and computer engineers to tell the differences" work that complicates this picture, that's still worthwhile to look for and into.)
Oct 21, 2026, 8:32 PM@gaditb [footnote: "The price of civilization is eternal maintenance" is a verbatim quote from somewhere in the middle of a web serial I read years ago in the depths of burnout and can no longer find. It wasn't a very good web serial except for that one line.]
Oct 21, 2026, 8:37 PM@zwol (I love love love finding mediocre- or bad- (or even just uneven-) -quality media and enjoying it as a just "but they CARED when making it, and I enjoy watching them care", and finding in one corner of it the one or two places where -- for this line, for this scene, for this moment -- OOH, that hits. beautifully, truely, deeply, passionately, small but as strongly as a piece of a critically-acclaimed work does. Just, accumulating those fragments of beauty.)
Oct 21, 2026, 8:44 PM@zwol But "Writing a program that works for sure, always, for everyone" should be the definition of programming, at least (semi-)professionally. I sing in my shower, with good pitch, I can sing in a karaoke, with success, but am I a singer? No! Real singers have voice techniques and train for hours so they're able to sing and perform in any circumstance, at the drop of a hat, whatever the weather. I can't do that, and it's okay, because I'm not putting in the work to do that.
Programming is the only "engineering" discipline where we somehow think it's okay to deliver something that works approximately in the common case. Would you live in a house made by an amateur Sunday architect? Would you fly in a plane made by a dilettante avionics engineer? Why do we accept from the programming discipline what we do not accept from anything else? We often say "because the consequences aren't as dire", but is it true? Haven't lives been lost because of bad programming? If not lives, how much money has been set on fire because of bugs and/or security holes?
So when you say "Programming isn't all that difficult", I disagree. Being a script kiddie or a wannabe programmer isn't all that difficult. But programming includes the enormous amount of fiddly, tedious detail work that you're talking about, and it is because we as a society fail to recognize this that "software is bad". Yeah, any human construction would be bad if they were also made by DIYers who never properly learned a discipline that was never properly formalized or secured or regulated.
Oct 21, 2026, 8:47 PM@ska I see where you're coming from, but I disagree with your framing for pedagogical reasons. I think it's important to be clear that "write a program that kinda works for you" *is programming* and is in fact not that hard, in the same way that "build a trestle table that will hold up for what you need it for" is not a terribly difficult carpentry project.
Because "computers are incomprehensible to everyone but the highly trained priesthood" is a _worse_ meme to have in the zeitgeist.
Oct 21, 2026, 8:54 PM@zwol I understand the pedagogy reasons, and agree that we should not make programming more hermetic than it already is, but 1. two wrongs don't make a right and 2. rigor isn't necessarily hermetism.
I want everyone to be able to build a table that will hold up for its purposes. But when we're talking about building houses, or even the timber that will be used in houses, it might not be a bad idea to be a little more rigorous.
Oct 21, 2026, 8:48 PM@zwol Crud, I think this even made it into the 1995 edition of The Mythical Man Month... it's not new, and yet every time it comes up it's news.