This is interesting, I was aware of George Hotz back when he did create that self driving setup of his using AI, it kind of showed up Tesla’s self driving system. Guy is brilliant. But I think this is the difference between a hacker vs a developer, and I think this guy is a hacker-type.
A hacker is brilliant but doesn’t work well with others or doing modular work or focus heavily on standards and practices. They do good work to get to their goal, it may be a hacked up mess but it works and works well (usually). A team of developers would do the same work towards the same goal but take a lot longer. But it’d be documented and use standards and have been tested, extensively, etc.
So I guess Twitter has taken to hiring hacker-types in short bursts to get some needed changes implemented. I suspect he’ll succeed but I wonder if the code will be better afterwards, or just a patched hacked up mess on top of the existing code. I wonder how documented his modifications will be and how easy it’ll be to maintain and add on to in the future.
Still, if he’s as good as I suspect he is, he will write good code, it will do good things for the changes he wants to make to Twitter, and it’ll look good and work. But not sure how messy it will be on the backend.
hehe, maybe that’s the way to go… hire hacker-type of people to develop some new software or feature or whatever on a short term basis. They’ll work their rear off getting it done, because that’s just how they work, and they’ll get it kicked out in 1/4 the time it’d take developers to do it. Then you and the hacker-type part ways on good terms and now you release the product or feature, etc. Then turn around and use your team of developers to reverse engineer the guys code, clean it up, document it, make it more standardized, etc. spend the next few months molding that code into more acceptable “corporate” standards and then you can begin releasing updates to the product/service/software and bug fixes, etc. once your developers have rebuilt it and understand it.
Basically have somebody build a brilliant but crude prototype that works, and do it quickly, and if there’s an immediate need you can even release that prototype as production (if it’s good enough). But then spend time learning your own prototype, reverse engineering it, cleaning it up, and later enhancing it.
- @ 2022-11-23 4:04 pm