from HaraldvonBlauzahn@feddit.org to programming@programming.dev on 27 Mar 06:12
https://feddit.org/post/27682092
[…] Coding agents are now also introduced to production codebases. After 12 months, we are now beginning to see the effects of all that “progress”. Here’s my current view.
[…]
All of this compounds into an unrecoverable mess of complexity. The exact same mess you find in human-made enterprise codebases. Those arrive at that state because the pain is distributed over a massive amount of people. The individual suffering doesn’t pass the threshold of “I need to fix this”. The individual might not even have the means to fix things. And organizations have super high pain tolerance. But human-made enterprise codebases take years to get there. The organization slowly evolves along with the complexity in a demented kind of synergy and learns how to deal with it.
With agents and a team of 2 humans, you can get to that complexity within weeks.
#programming
threaded - newest
That rings so true.
But the thing is, pain is a warning signal. If you go jogging completely drunk and hit a tree with 6 mph, it will be painful, yes, but the pain will warn you not to do it again.
But what if you move 12 times faster?
If you drive a light motorcycle completely drunk and with no helmet, and hit a tree, pain will not be able to save you.
For company legacy codebases, yes they are dysfunctional but they have found a kind of precarious equilibrium in so far as they exist because they are making money and thus are useful by some metric. The slow movement and requirement to work somehow balances the unstoppable (with in company practices) growth of entropy and messiness.
And in a way, the money is an analgetic for the pain. Or more sharply, big companies act like junkies on a money drug because money is the only thing that ever counts.
Figure what happens if entropy is grown 100 times faster…
if you don’t know how to make good code you can’t expect AI to do it either
if you can code well then AI can be a great help
This whole ordeal is going to be quite a wake-up call for some orgs. They spent this whole time thinking they only paid devs to produce code, not to understand it, because the understanding part was invisible to them.
Now they’re like astronauts blasting off to space and not knowing they need to bring oxygen with them, because they’re never seen any.
You missed the most important sentence… “While all of this is anecdotal”. The rest you should read as though you are talking to the ceo of nvidia after he bought a huge short position on every AI company in the world.
You can do a task pretty well if you nudge the AI, have it write an exact explanation about every part of the architecture, code and data flow it’s working with and throw relevant files into context, and correct anything that’s wrong before you send it to do the task. You still have to review, but I didn’t have to correct much in my experience.
But that burns like 20$ of tokens per task, at current prices that are way below the costs AI companies are paying.
While it does help me, especially with parts of the codebase I’m not familliar with, it’s not sustainable, and it’s actively and very quickly robbing me of my skills and knowledge. It’s really a bad idea to use it, in two years time you’ll be royally fucked once they raise prices to recover the trillions they are loosing right now.
So, however tempting, I simply don’t use it. I won’t throw away years of college and experience just to do a task a little bit faster today.
How are you learning a new tech stack, if the agent is doing it all for you?