Hacker News new | ask | show | jobs
by programminggeek 2382 days ago
This model is sort of absurd. It doesn't take into account the most important factor... people.

If you made a model of the world today back in 1920, it would be comical.

Imagine you went back to say 1920 and told people that in 2020 they would be surrounded by talking televisions, phones, that we'd be sending people and ships to the moon, mars, etc. and that they could instantly talk to any person on the planet (including loved ones on the other side of the world), they would not believe you.

In the last 100 years (heck even the last 20-30) we've invented a staggering amount of new and interesting things.

In the last 100 years people created interstate highway systems, nuclear power, the internet, radio data networks, self guided weapons, self guided vehicles, robots, rudimentary AI, etc...

What if in 100 years we make fusion work (probably for space travel/war?) and then we replace the existing infrastructure with that. And what if we can do enough carbon removal or storage to solve this?

Any model that doesn't model the history altering inventiveness of humans is not an accurate model at all. It's just really fancy navel gazing.

1 comments

You can't model things like innovation on a hundred year time scale. The reason the predictions from the 1920s "would be comical" is not because people in the 1920s where all fools but because modeling future innovation over a 100 year time scale is extremely challenging. Even if you could include innovation it would reduce the benefit of such a model. Imagine you have a model that has access to a global temperature oracle and it says: "global temperature decreased 0.1 C in 2120 because of innovation". You don't know why that happened and so such a model would be mostly useless to achieving any result.

What you are proposing is like a software performance analytics tool that already includes the fact that you will improve the software performance in the current metrics. However the purpose of such a tool is to figure out where the performance problems are so that you can fix them. Adding such a feature makes the tool useless. What you want to see the baseline and then you want to model the impact of solutions after you discover them.

>What if in 100 years we make fusion work (probably for space travel/war?) and then we replace the existing infrastructure with that.

I think fusion will happen faster than that. However the model we are discussing only goes out 81 years so it doesn't cover the impact of events 100 years hence.