Computation as a service

DNA sequencing every big crop. But every agricultural big name is doing that now.

Sequence rich people's bloodlines? That's already being done.

Sequence populations by race? I can't see who would pay you for it.

You could rent your equipment out to low end schools and low end research organizations.

Or you could rent out your equipment to small scale medical startups near Stanford.

>>1593018

>DNA sequencing

How marginal of an improvement would a 100% accurate method be over what the big names are currently doing?

How would you monetize the improvement or incentivize these big names to rely on your service?

>equipment

While I suppose the scope of the question might include a method that relies on specific equipment, let's say that you have a monopoly on a computational method. That is: you can answer a question nobody else can (or, at least, you can answer a specific question better than anybody else can).

>>1593014
SEO.

Even a 5 year old could figure out how to make multiple billions if you could guarantee them being the top result on Google, especially for unrelated search terms.

>>1593030

Interesting. So how do you pose SEO as a computational problem? How would you specify that problem to the service?

One issue I see with the service is that there are can only be so many top results for any one query; it doesn't seem to intuitively be a computational problem to induce some effect on a system with closed (not publicly known) parameters like a search engine.

>>1593039
>So how do you pose SEO as a computational problem?

How is it not? Page ranking is determined by algorithms. If you've got a good way to solve for those algorithms, you can create a model to build a successful site and try test cases without wasting time publishing them. From that you can build up a simplified model that can simply generate websites on the fly.

Unless you mean specific code examples, in which case, I don't know. You merely asked what I'd solve if I could solve one thing perfectly, for money, and how to monetize it. That's easy. Get to page one on google, and collect revenue from advertisements, scams, and malware.

If you want to speak about actual code, I'd suggest asking the spambots generating clickbait and rehosting news articles surrounded by hundreds of ads and malware 'security scanners'. It's already a fucking enormous industry. Hundreds of billions of dollars. Selling a majority stake in that market to someone at any price you could dream of is easier than selling meth to meth-heads.

>>1593014

This question is retarded because the issue is not with access to computational power, but with how to effectively apply it. Anything that you can do is inherently hamstrung by the human implementation. In other words, a human has to design the algorithm and application of the program, and therein lies the problem.

>>1593057

Assume that this method may be easily implemented. The point here is that you have a monoply on a method that can be effectively turned into a service.

>>1593053

The issue with SEO as a computational problem is that the precise algorithm used for ranking is known. Because they are operated privately, there is also an issue of human 'tuning' beyond the algorithm itself. From an information-theoretic perspective, such a system is not computationally distinguishable or predictable.

>>1593065

is not known**

>>1593065
>The issue with SEO as a computational problem is that the precise algorithm used for ranking is not known.

>To further clarify things, let's say that this method is (1) 100% correct
>all existing alternatives being approximations

Your OP already assumes you are solving an 'unknown' arbitrary problem.

>>1593065

Then the answer is market/sentiment analysis (particularly futures).