-
I’m not a software engineer, but…
If the issue is ‘10 years ago we bodged something, now our whole system rests on it, but we can’t fix it because it costs too much’
Surely that cost is only gonna come down? Certainly in terms of time required.
Obviously AI isn’t a one click fix, but whether it’s finding the issues, fixing them, or testing them, it’s gonna speed it up?
steelspeed-
Obviously AI isn’t a one click fix, but whether it’s finding the issues, fixing them, or testing them, it’s gonna speed it up?
It will speed things up, but it speeds up whoever is using it. If the person using it doesn't have the experience and judgement to know when it's not giving the best advice then they're going to trust it blindly and just make a mess of things even faster.
A shit analogy would be: powertools don't make everyone amazing at DIY, but they do enable the bodgers to make the same mistakes faster. Or make the people that wouldn't tackle a DIY project at all likely to have a go (and make a hash of it).
-
Surely that cost is only gonna come down? Certainly in terms of time required.
Systems almost never get smaller, but keep accreting new features.
So, typically the cost is only gonna go up, as more stuff comes to depend on the bodge and the precise details of all those dependencies are slowly forgotten or mis-remembered.And while you could use AI to fix individual bits of code, that's never the bottleneck. You can't use it to analyse all the dependencies & their partially-documented functional requirements, plan and run integration tests, or figure out how to manage upgrades around whatever else is going on, and it won't help you figure out who needs to agree and get them all in a meeting.
Greenbank
useless
Generally? Other than the digital tech revolution slowing down rather than exponentially expanding, there is none.
@steelspeed
No. Mitigating technical debt requires understanding and insight. It's not about bugs. Most things labelled "technical debt" are entirely functional, just not a solution you want to have to live with long term.
Technical debt is not a bad thing in itself; the metaphor is useful because financial debt is something that people and companies thrive on, as long as it's acknowledged, logged and managed. Financial debt is bad when people hide it and/or don't keep it in check. Technical debt the same. The software/IT world has such a bad problem with unmanaged technical debt mostly because of the pressures I mentioned above, partly because most people (and managers) don't have the same basic understanding of tech that they do of their account balance. It's not hard to get people to see the shakey step ladder problem when you're actually talking about building a house. Trying to get them to see the real software engineering equivalents, and why they'll be a problem in the long term, is hard.