[sheepmullet]

> but economists tend to agree that my outcome is more likely than yours

Would you play a round of Russian Roulette for $100? The most likely outcome is positive.

> Should we ban all software development just to be safe....

Why do you keep trying to reduce the argument to the absurd?

I haven’t made any mention of banning immigration and yet you are contrasting my position to banning software development... like somebody working on a web browser will magically invent a new master AI.

[learc83]

>Would you play a round of Russian Roulette for $100? The most likely outcome is positive.

>Why do you keep trying to reduce the argument to the absurd?

Really?

In your Russian Roulette example, we know the risk, we know the rewards, and we know the probabilities of both.

With regards to immigration, you believe you can't estimate the probabilities, so you've arbitrarily decided that expected the risk outweighs the expected reward.

>I haven’t made any mention of banning immigration and yet you are contrasting my position to banning software development... like somebody working on a web browser will magically invent a new master AI.

So if that's the problem you have with the argument I’ll change it to just banning just AI research?

Let's go another route. Automation through software development has the potential to cause very similar problems to what you think immigration will cause--depressed wages, high unemployment etc…

There’s a big potential downside, we can’t realistically undo it once it’s done. We don’t know the probabilities because the experts can't be trusted. The sheepmullet doctrine says we should ban it (or severely restrict it?).

[sheepmullet]

> With regards to immigration, you believe you can't estimate the probabilities, so you've arbitrarily decided that expected the risk outweighs the expected reward.

Think about what you are saying.

What probability of serious harm are you willing to accept in return for a half percentage point pay rise?

You must be certain the chance of things going wrong is almost nil.

What gives you such certainty? Your economics experts certainly aren’t claiming to offer such certainty.

> Automation through software development has the potential to cause very similar problems

At the end of the day it’s a significantly easier problem to manage.

Software can’t vote, get welfare, etc. Software doesn’t have any rights.

And of course even on this much simpler problem we should still apply risk management techniques.

[learc83]

Just using the term "risk management techniques" to make your position sound more credible, doesn't actually make it more credible.

Again, you're completely ignoring the potential negative consequences of severely restricted immigration, and understating the likely positive benefits.