To democratize social media you need a market of “algorithms” (to use
the common meaning of the word, ie. a set of rules deciding which
content to show you).
This would use, among other things, natural language classifiers which
could filter and rank content by topic, relevance, and other
interesting attributes.
Such market would be a great incentive for people to build better
algorithms, while not keeping the overall system under centralized
control. If it gets too corrupt or undemocratic, yoink, it gets
forked.
How can we make each classifier be a little securitized business?
Since it would produce labels, it would be producing value. It would
have intrinsic value, like a farm, or a cow. (Cow token? perhaps
Buffett wouldn’t hate it) One could buy shares in a classifier and
earn dividends.
This could be done with security tokens (or perhaps NFTs? Which would
be better?)
The big missing puzzle piece here is: I’d like to keep the source code
of the classifier (and perhaps, its training data) secret unless you are
the majority shareholder of the asset. Can this last part be done?
PS. Perhaps the thing to securitize is the farm, not the cow. They are more likely to build long term moats, or temporary monopolies (to use Thiel’s term) than a single classifier or algorithm would. OK so now this involves securitizing a “business” that produces these algorithms / classifiers, which must be a problem that has been addressed or at least thought of in the Ethereum community. Is this the case? There must be tools for this in the Ethereum ecosystem, right?
In any case, for this specific application, the “business” could just be the “admins” of a security that owns several classifiers, each of which produces revenues, and has some costs. Then in meatspace, an actual business could own that digital asset.
It seems like I may not have posted in the ideal location for my question. If anyone knows some other place online where I could hear feedback on this idea I would very much appreciate it. Thanks in advance!
Thanks Josh, good point, I thought about that but I really think that upon contact with the practical world, especially in the early stages of decentralization, inevitably you can’t make everything ideally decentralized. Which is not to say this wouldn’t be a beneficial thing to do.
I did try reddit, too. I’d love to hear about any devil-in-the-details issues with this general idea, any misconceptions I may be under, or any other criticisms.
I’d build “clients” on this network, so I see this as something that could be a worthwhile “public good” to build.
The biggest hesitation I’ve had is that it seems like if you want a long term productive asset to invest in, the natural unit is not the single machine learning algo, data, or combination thereof, it’s the institution, the company, that creates and improves it.
So are we back to square one? Does this mean that current financial systems solve the problem/opportunity I’m describing?