Artificial incompetence

Is Github Copilot something to get your panties in a twist over? Well, no, not really. But what’s more interesting is what made you think it would be?

I’ve been developing software for somethere north of 35 years. It has been so long that I’ve forgotten precisely how long it has been so I probably give a different answer every time someone asks.

This doesn’t necessarily make me a better software engineer, it just means that I’ve seen a lot of stuff before and it gets really hard to get enthusiastic about new things - because they are almost never new ideas.

Think of something “new” you have learnt or discovered in the last decade and there is a good chance that it is something that was already an established practice somewhere. There are really very few original ideas. If you think carefully about it, it is highly unlikely that you have had a single original idea of note your entire life. Someone else has already been there.

But don’t feel bad; ideas aren’t actually all that important. What matters is if you can turn an idea into something that is of use. And this is a good thing because it means that even though you are not likely to ever think an original thought for your entire existence, you can still be an innovator.

Github Copilot

It was inevitable that someone would sooner or later produce something like Github Copilot. I mean, all the pieces were part of the lexicon of a large population of programmers. Sooner or later someone would plug this doodad into that doohickey and make something that supposedly helps you write code.

And it was as inevitable that people would get really enthusiastic about it and extrapolate wildly to where the programmer becomes an extinct species. If you are an old fart like me, you have probably seen this 3-4 times already in your career. A bunch of content, something that is supposedly artificial intelligence and some stimulus provided by interaction and then some magic symbiosis. Yes, it is a novel application, but it is only exciting if it works unreasonably well.

But so far, it doesn’t. It’s kind of meh.

Yes, it is an interesting experiment and sure, I think the idea has some merit. I’m sure it will evolve into something useful. But it isn’t earth-shattering. It doesn’t make you explosively productive. It isn’t exactly the bicycle for the mind people want to believe it is.

At least not yet.

And here is the thing: while the idea is good’ish, it is really what you can do with it that matters. There is still stuff you can do with this.

I have some highly creative friends who disagree violently with me when I say ideas do not matter - only execution. Ironically: I have yet to identify any of them actually having a single original idea between them, and as far as I can see, the thing that separates them from the rest is their ability to execute - not originality. I’ll grant them this though: this presupposes that people are well-read and are aware of ideas - a state of affairs which is in decline.

If I’m not excited about an idea, be it yours or someone else’s, it is mainly because it needs two things to be worth getting excited about:

• You must implement it and implement it well.
• It must provide at least one order of magnitude improvement along one or more dimensions.

If none of those criteria are fulfilled, then it is just an idea. It doesn’t actually change anything.