3 ms·
This kind of complex-system-optimization problem seems ripe for improvement via machine-learning. Cases where 1) precisely modeling the outcome is totally intra
by brundolf 6y ago
This kind of complex-system-optimization problem seems ripe for improvement via machine-learning. Cases where 1) precisely modeling the outcome is totally intractable, and 2) you really just need to get "close enough" to get good results. It also helps that this case is predictive in nature (so it's easy to go back and measure accuracy) and has an enormous set of available training data (all traffic happening everywhere across the country).
It feels very similar in spirit to that story about Google reducing data center energy use by 40% via ML optimization: https://deepmind.com/blog/article/deepmind-ai-reduces-google-data-centre-cooling-bill-40 https://deepmind.com/blog/article/deepmind-ai-reduces-google...
More broadly I've started wondering if there's a huge opportunity to apply software to civil improvement. Everything from detecting patterns of corruption, to aggregating data that can inform voters, to identifying sources of inefficient spending. I don't put much stock in things like digital voting/governance, but in the area of "data crunching and presentation" I think there's a lot that could be done.
- nicolaskruchten 6y ago> an enormous set of available training data (all traffic happening everywhere across the country). I'm not sure this is very helpful to predict traffic patterns over the next 20-30 years in a policy-sensitive way... All the historical data in the world wouldn't have predicted the rise of Uber, the pandemic, the maybe-eventual rise of self-driving cars etc.
- brundolf 6y agoTrue, but it doesn't sound like the approach described in the article factors in extraordinary circumstances like that either
- nicolaskruchten 6y agoThe model won't tell you how likely these scenarios are, but one advantage of this class of model over black-box ML function-fitting is that you can do scenario planning. You can do models runs like "what happens if the price of oil shoots way up?" or "what happens if 50% of the cars are taxis?" (e.g. Uber-driven future) or "what happens if transit all of a sudden gains a huge perceived cost?" (e.g. respiratory-virus pandemic) in the third and fourth stages of the model. In the early stages of the model you can look at stuff like "what happens if firms abandon city centers?" or "what happens if 50% of white-collar work is done from home?" etc.