Once a week would be enough, assuming weekly distribution of TheCoin.  Is a 3-5 minute walk once a week really that much to ask?   If someone thinks it is, maybe they don't need TheCoin distributions.   (Daily position checking was just an attempt to make life harder for those who will hire poorer people to collect TheCoin for them.  But on reflection, they'll mostly hire people who don't yet have their own wireless internet device, and it's better that those poor coin farmers get some income, than none.)


Honestly - I've previously considered and rejected a number of the most obvious options, including a real world verification organization. 

So far the unique spatial position test, backed by random real-time interactions with others who are also verifying their positional identity, is so far the simplest and quickest and best.  Yes, it still has flaws.  Not everyone has mobile internet (yet).  Until everyone can afford that, crooks can hire others to walk for them.  Not everyone could physically perform the "walk".  (Perhaps have randomly selected people verify the physical location of the disabled person as part of their own walks.)   

Hmmm - new names for TheCoin:  "Walkers".  Or  "Walkies".   
-----

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.   In areas with thin or corrupt law enforcement, crooks would take over the local organizations.  Desire to maintain a good reputation is simply not a strong enough motivator - as evidenced by the many known instances of corrupt behavior in organizations, plus the larger number that go undetected.
-----

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".   

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.