Hacker News Reader

Comment by harryh | Hacker News Reader

If you look at what most people were doing in 1930 most of those jobs are obsolete. And yet most people are still working. I'd be pretty surprised if 2015-2100 turns out to be much different from 1930 - 2015.

View on HN

That assumes that the automation trends of the next 100 years are going to look anything at all like the automation trends of the last 100 years. I don't think that's likely due to pretty obvious technological multipliers. Progress isn't linear.

Additionally, I think if you spend much time around the working classes, even today, you'll find that they're living in a world that's leaving them behind in a way that simply wasn't true even 30 years ago.

Also, "most people are still working" is increasingly less true:

* http://www.forbes.com/sites/toddganos/2012/11/03/what-is-rea...

A significantly greater percentage of the working age population is working at a job today than 85 years ago.

Yes, that's true, but that's almost certainly because more women work now than ever did before. That's a one-time (enormous) increase, but there's no reason to believe it's an indication of a longer-term trend. In fact, the data show otherwise.

You can play around with the data here:

* http://data.bls.gov/timeseries/LNS12300000

I am unmoved by the relatively tiny short term movements you are referring to. Basically all you have shown is that we had a relatively large recession in 2007-08.

The past doesn't necessarily predict the future. I really suspect that it might end up being different this time.

My working hypothesis is that software-defined Turing-complete automation is qualitatively different from special-case automation. It permits the rapid reconfiguration of general purpose machines with software, which is cheap and fast to produce in comparison with pretty much any physical system. Combined with simulation it allows for very rapid iteration.

You can in many cases effectively design something and then just hit 'print.' Some human intervention and assembly and some custom work is usually required, but the quantity is far lower and the skill level far higher than in the past. In the end you need a small number of highly skilled employees, not a large number of low-skill ones. The conventional 'worker' is obsolete.

Compare one-off factory tooling set for stamping out one part with a 3d printer, for example. The 3d printer is what the future looks like.

The alternative to a universal basic income here would be something like the world depicted in The Hunger Games series -- tiny super-rich walled totalitarian city-states presiding over a majority that has regressed to 19th century or prior standards of living. I'd personally consider this to be a new dark age.

Yes, perhaps THIS TIME IT'S DIFFERENT. Perhaps.

I'm just saying I'd be surprised.

¯\_(ツ)_/¯

If it were possible to predict the future on the basis of the past, everyone would always make money in the stock market.

Over the long run most people have always made money in the stock market.

Are you making the simple claim that, over say a 30 year period, certain broad stock market indexes ended higher than they started? Or are you making a much more interesting claim? I'd like to see the data set for that if you can point me in the right direction.

I'm making the simple claim.

This ain't your grand-daddy's automation.

What I think what most people count on is the inanely fast ACCELERATION of technological progress. The amount of advancement between 1930 and 2015 will be orders of magnitude less than 2015-2100.

Why do you say that? I think the technological progress will continue to grow exponentially. There are more researchers now than at any point in human history. The economic incentives are still at least as big (relatively speaking).

At the moment there are many industries with great potential for applying machine learning and various other automation techniques.

And considering that many leading AI scientists are expecting smarter-than-human AI within this century (or the start of the next one), I certainly fail to see your point.

Anyway, there is still a lot of room for exponential growth of human knowledge, in a lot of fields.

I think you misread what he said.

Reply on news.ycombinator.com