TheCoin success depends on rapid roll-out. Building up real world ID verification organizations would take too long - likely a decade to even get coverage of most major cities on Earth, decades to get close to universal coverage, even assuming no political interference.
Then you're skrewed. Rolling this out will take beyond decades, just to marshall the resources that would be necessary to ensure that 90+% of the world's population have access to some device, power, signal, markets, goods, et cetera. Plus, the environmentalist wackos will fight you every step of the way--they don't seem to realize that 1st World countries are not going to give up their standard of living, which means significantly more resources (coal and oil 'cuz dey hate dat nuqulear stuff) will be expended to bring the entire population up to even a modicum of bare necessities.
Biometrics sound good - but unique personal attributes are subject to high measurement error/variation. A person could count on a slightly different digital representation from every capture. Hash that, and you'll get a different hash every time - allowing multiple identity creation. Attempts to eliminate those errors reduce the uniqueness of the measurements - creating higher likelihood of more than one person with the same ID code.
Consider DNA sequencing (though it will be too expensive, for several more decades, to be a practical global solution): Even the best (expensive) sequencing has error rates of 1 in a million bases. ("Cheap" methods, ~1 in 1000.) For the ~3 billion bases in the human genome that's at least ~3000 errors per sequencing. No two sequencings of a person's DNA are likely to ever give the same digital representation.
One might imagine cleverly "averaging out" DNA sequencing errors somehow. But human DNA is 99.9% the same for everyone. Blood relatives will be even more similar - perhaps only 1 in a million bases differing. That's on the same order of magnitude of the best sequencing error. Trying to eliminate the errors to get a repeatable digital representation, will map more than one person to the same digital representation, and hence to the same hash code "identity".
This may be true now, but in the tens of decades it will take you to ever start setting The Plan, that won't be the case. Especially as I didn't really mean just DNA. It could just use all of the following: DNA, fingerprints, retinal scan, finger length ratios, et cetera. You'll be able to get the error/cheat rate to essentially zero.
CheapID is interesting for the privacy protection it aims to grant - but still subject to easy acquisition of fake IDs, as well as corruption of accreditation agencies.
I tried to come up with a solution based on Webs of Trust. But a WoT is not as good at preventing a person from having multiple keys by getting different groups to verify trust in them. Casual cheaters could probably be blocked by requiring them to inform their trust networks of their name, address, etc. But a small group of cheaters could build their own sub-web of trust with a number of fake identities per real person. The best counter I came up with for that was bounties for catching cheaters. But that adds a lot of complexity for verification of cheater reports.
Yep. Cost after cost after cost after cost. And you still haven't gotten to the earnings/mining issues.