Relevant: https://twitter.com/ConsultingHumor
We also serve as convenient scape goats in case risky projects go bad.
AbstractBoondoggle boondoggle = new AbstractBoondoggleFactory()As the length of a thread about software design increases, the probability of someone mentioning a "general-purpose tool-building factory factory factory" approaches one.
Including the act of quoting Thor's Law.
———
Though in reality does that mean that the guy bought a metal 3D printer?
To me, ChatGPT's output is indistinguishable from what I've seen offshored coders produce. Only difference is you get the results in 30 seconds in a chat window instead of overnight in a zip file emailed to you.
There are whole businesses here in America that only do rewrites for offshored projects.
After getting ChatGPT to write some code for me, this could happen again except with AI instead of an outsourced team.
So far, it understands the requirements better and writes pretty good code on average. Its unit tests are pretty good too. It always requires some fixes and integration, of course.
This included lots of offshoring companies, like IBM, that seemed more about filling out whatever the budget sheet given to them than actually making software or even any evidence of software (commits, issues, etc).
I expect that there are firms out there right now using ChatGPT to produce the facade of work so instead of messing around and billing 8 hours times 30 people every day, they just have a few people fiddling with prompts and checking in the generated text.
I actually think this might have the same impact but would hope that the value just goes back to the client in having employees do the AI work directly.
The joke was that behind every successful offshore project was one or two really busy onshore engineers manically covering and integrating work from the away team.