A friend sent me a link to this - nice work, and I happen to be intermittently working on a very similar (and unfortunately similarly named) project - https://github.com/jarfa/ML_from_scratch/. Check my commit history if you suspect me of copying you ;)

I don't think I'll be implementing as many algorithms as you though, I should force myself to work on more projects outside my comfort zone.

Oh, cool! Haha, that's unfortunate. Maybe I should have done a better job finding an original name. Good luck to you anyways!

The Midwest has many large cities.

But the Midwest does not completely overlap the wind belt. The more densely populated parts of the Midwest are especially far away.

Right, and places like Texas, Colorado, Oklahoma, and Montana are not included in the Midwest. The point isn't that the wind belt region makes up exactly 1/5 of the population, but that there are many people there that shouldn't be dismissed because they aren't in the most densely populated cities.

Oklahoma is definitely included in this RTO's footprint as is Kansas and parts of Montana.

The wind farms in SW Minnesota/the eastern Dakotas/ NW Iowa (http://www.nrel.gov/gis/images/30m_US_Wind.jpg) are relatively close to the larger urban areas in the region and I suspect some of it might even feed into Chicago.

There's not much overlap between the wind belt and where people live. But there are many round farms in the wind belt. The unused corners of round farms would be good places for windmills. There's road access in place. Extra revenue for farmers. Won't bother the crops.

Collecting up the power and shipping it somewhere useful is a problem. That requires new high voltage transmission lines and collection at substations for conversion to DC and long distance transmission.

This isn't great news for those who care about global warming, nothing drives cleaner fuel technologies better than high hydrocarbon prices.

This isn't great news for those who 'don't care' about global warming - people hungry for fossil fuel jobs, either.

It's becoming less and less obvious that anyone in a position of power advocating for fossil fuel energy actually cares about the jobs.

It was always rather transparent that the "jobs" angle was a play by big oil to stoke populism for an objectively unpopular agenda (i.e. more drilling, extraction pollution). Once automation removes the need for low skilled laborers I would think this tactic would not be available anymore (but who knows, in this era of alternative facts).

Lots of jobs in gas and oil are very highly skilled jobs. Automating the stuff that's high paying because it's risky, dirty, and far from the workers' homes will increase pay imbalances as the software developers, automation folks, sysadmins, robotics engineers, mechanical engineers, petroleum engineers, geologists, executives, et al will continue to get their pay and probably more while anyone automated away who doesn't retrain will make far less in another role.

Well, of course. Jobs are a cost center - and the people advocating hardest for fossil fuels are resource companies.

The fewer jobs they're creating, the less pull they have with congress...

That's the useful fiction. Reality: Dollars contributed to campaigns.

I've never been able to get a clear picture of how dollars translate to votes.

Especially given the prevalence of "safe districts" where one party has very solid footing, it seems hard to believe that campaign funding is all that important.

This is very interesting, I'm curious as to how this works outside a controlled environment like the CIA.

What sort of arbitration process do you have for judging the quality of evidence and the degree to which evidence agrees with hypotheses?

What's to stop trolls from spoiling the process through the above methods, or through adding extraneous hypotheses to confuse the process?

Thanks for your interest! Adapting the technique for the internet is one of the main challenges the project is trying to address. That's the reason the main site is invite-only right now.

Here are some thoughts on the points you bring up:

- You can currently dig into the evidence quality by looking at the corroborating/conflicting sources that users have added. Users can also tag sources based on their quality, e.g., state-sponsored media, secondhand information, etc. Eventually I want to arrive at some metrics for evidence/source quality.

- The degree to which the evidence agrees with the hypotheses is handled by the evaluation/assessment process. The site merges the evaluation/assessments from all the participants and highlights areas of dispute. In the future, we'll probably end up allowing users to mark which evaluators they trust (e.g., verified journalists).

- As far as trolls go, we're exploring the space of moderation and anti-troll measures. This will include automated measures like rate limiting, as well as moderation features like flagging evidence/hypotheses as irrelevant, duplicate, etc.

A lot of potential ideas are captured on the issue tracker: https://github.com/twschiller/open-synthesis/issues