I think we see the results of rushing in all the software around us much more than we see the delays of perfectionism.
I think we see the results of rushing in all the software around us much more than we see the delays of perfectionism.
Thankfully the market correction made this much less of an issue.
I practice my perfectionism at home, on hobby projects, where I can run lint with every option if I want, and resolve every warning, static analysis issue, style issue, I can profile my code and find/fix slow parts. I can run valgrind on it and find memory leaks. And I never have to release the software! None of these things were common practice in any software job I've ever had. It was always "Get it to barely work and then ship ship ship!"
Maybe it's the pessimism of age, but it feels like that describes all biology, economies, politics, etc...
While the article doesn't offer specific examples, it's probably talking about this sort of stuff. e.g. "perfecting code, often code that hadn’t been touched in years and didn’t need to be touched". This sounds like a classic "oh, this isn't how I would have written it, so let me just refactor everything here".
"Not how I would have written it" does not equal "bad".
This isn't making the code better, less bug-free, or "more perfect" in any way. Actually, it often introduces bugs.
I'll take a 95% solution delivered in 2 weeks over a 99% one delivered in 2 months.
This depends a lot on the context. If I deliver the 95% in 2 weeks can I ship an update in 2 months that delivers the 99%, or are we stuck forever at 95% (due to technical limitations or business reality)?
If I'll be able to deliver the 99% soon, then sure, let's ship the 95% sooner rather than later.
But if for whatever reason I realistically cannot do that, then you have to consider the context. What is left out in the 95%? How soon will it be a problem that we're missing that remaining 4%? Who will notice the missing pieces? Will they be able to work around their absence?
Many times you may still come to the conclusion that the 95% is the right move, but sometimes it's better to not offer the two-week plan as an option if the fallout from that missing 4% is going to be high.
E.g., the Twitch codebase
They can't change it or add new features or fix problems because it's too much of an unholy startup crack-fueled mess.
Not even companies who stole the leaked codebase can improve the horrible user experience on their Twitch clones whatsoever.