Hacker News Reader

Comment by plainOldText | Hacker News Reader

I think these are key thoughts worth considering going forward:

  > Code is not an asset, it's a liability.

  > Every line must be maintained, debugged, secured, and eventually replaced. The real asset is the business capability that code enables.

  > The skill that survives and thrives isn't writing code. It's architecting systems. And that's the one thing AI can't do.

View on HN

Code is an asset. It costs a lot of money to create “code”.

This is like saying your transportation fleet as a delivery company isn’t an asset but a liability, it makes no sense.

Almost all assets require varying amounts of maintenance.

I imagine a factory with giant containers of hideously corrosive and toxic chemicals that might be very useful for a particular chemical reaction.

Those default ground-state of those tanks are liabilities. They may be assets today, but maintaining that conditional status requires constant work.

The thing is, as compute prices fall and AI advances in its capability, swarms of coding agents can sweat on every line of code, looking for cleaner implementations, testing them, 24hrs a day. Current models often produce spaghetti crap. But it's a bold assumption that will always be the case.

There's a pretty big herd of sacred cows in programming and this debate always surfaces them. But I remember similar arguments being made about Go once and its version of human beings' special something. We saw then that sacred cows don't live long when AI really arrives.

I don't think comparing Go to developing systems to be useful, Go has a very clear goal to be achieved, it can be easily iterated over to achieve that singular goal. Developing systems is a much more complex task involving a lot of human subjectivity to even define a goal, there's whole swaths of attempts in methodology, processes, tools that try to even achieve some kind of definition of a software development goal, and none so far have been good enough to automate this process.

It can happen that AI will get good enough to help with the human aspect of software development but using playing Go as an analogy doesn't really work.

You'll notice if you reread my comment that what I was comparing was the notion of the human player / developer's special something.

Why its bold? Bold is being so sure in predicting the future.

IDK, after almost 70 years of the software business, we can't even yet have package repositories that are not subject to supply-chain attacks.

SQL, with all its warts, has not been dethroned. HTML is still king.

We still have C/C++ at the heart of every OS and programming ecosystems.

That’s not something to be proud of.

Reply on news.ycombinator.com