This is cute, but I'm not sure why it has to be centralized, other than SMS message costs (probably trivial) and the lack of friendly software for retrieving, comparing, and organizing these SMS messages on the typical cell phone. I can foresee many disputes where Alice and Chuck remember that the bill was 136.89 and that BillMonk's records reflect a typo, but Dana agrees with Bob that it was 163.89. A more robust method may be that everybody sends everybody else at the dinner a message indicating their understanding of the bill and how it was split.

Also, given this example I wonder if their user interface favors certain social customs (e.g. dividing the tab evenly), making those relatively user friendly, over other customs or other circumstances (e.g. Alice only ordered a club soda and thus only owes Bob $5.99, whereas Chuck owes for a filet mignon with caviar for $159.46, or any other small loan of money where the custom is simply to pay the amount owed back).

I also recommend that they round down to the nearest dollar. Friends at a restaraunt (in wealthy countries) don't quibble over cents, nor even do acquaintences who want any sort of future relationship, for that matter. Sheesh.

Actually, I have strong doubts about the whole approach. People use the ability to remember and return favors like paying for a dinner out as a social signal to indicate and build trust. That's usually why one person pays in the first place, rather than the group splitting the tab right away. In this signaling "game", using artificial props (especially a central server which becomes "the" authority) may be "cheating".

OTOH, there may be quite a few people who don't carry cash any more and thus have lost the ability to split a tab right away, yet want to have dinner with semi-strangers without so much burden in keeping track of favors. BillMonk, or my decentralized version of it, might be very helpful for them. However if this is the case, and mental transaction costs don't swamp the whole effort, then at least some of these people must already in the habit of sending SMS or e-mail messages such purposes. Does anybody know of anybody doing this?

> I also recommend that they round down to the nearest dollar. Friends at a restaraunt (in wealthy countries) don't quibble over cents, nor even do acquaintences who want any sort of future relationship, for that matter. Sheesh.

You would think this might be the case but actually customs vary quite dramatically. In the country I am in now, the deal is everyone pays their own, precisely, and it is settled that way. The waiter turns up and deals with everyone, in turn. If you pay for anyone else you'll disrupt things.

In another country, it works thusly: the most senior guy leaves early and pays the entire tab. The next most senior guy then leaves, and picks up the new tab, entirely... and on and on.

In a third country, rotation of rounds is done on a schedule. In another country, rotation is done randomly. In another country, everyone pays an even share, unless it is wildly disparate in which case they pay nothing.....

Countries differ and us foreigners have to work it out :-) Also, in general I've found that the poorer the country, the less the quibble. Go figure!

> People use the ability to remember and return favors like paying for a dinner out as a social signal to indicate and build trust.

Right. Nothing is obvious in this game. In our flowers currency, we are slowly learning how it works and how it doesn't work - it all seems so obvious, but before we discovered the specialness of handing a person a flower, it certainly was anything but obvious.

Can favours be automated and centralised? I don't know, I don't see it yet, and I'm sitting here with a flower token in my hand which derived conceptually from a flowers contract that is issued on my servers and can be traded across the planet!

I would view Bill Monk as a special case of a PTR, public transaction repository. Glad to see, it actually is up and running unlike the PTR idea but it only accommodates a small subset of a general transaction repository, and it is furthermore bonded to a particular device and communication channel-- another small subset.

Why is it so hard for people to grasp the generalized model of transactions? An economic resource is anything of value, to anybody-- anything with utility in the mind of one or more persons. Transfers of resources are of at least three important types: reciprocated, unreciprocated (gifts), reciprocated within relationships infeasible to model (including longterm relations, obedience relations, etc.), and forced. For modelers, it is unnecessary to proceed to address the reciprocal half of the transfer if you haven't got the model correct, for the basic economic transfer.

An economic transfer can have many attributes or details but must contain an identifier of the thing transferred, a giver, and a recipient, and a date/time that gets you down to the point or interval when the flow happened.

In my opinion it is perfectly useless to proceed into any project to implement a general transaction repository until there is agreement on the meanings of the symbols in the database. The basic question can be ignored in small local tribes where everybody already knows both, the people and the things in the database. In that case the GTR is useless since, who the hell needs a computer? You could use little slips of paper, or a PIM or anything you want.

For a GTR to be useful you need a common contacts list and product list. In turn, the users must come to some agreement on the management of these two things so that they are not abused-- SOunds simple but has never been achieved. Every party list that has ever been created, has been exploited in the three usual ways that information is always exploited (blocking some information from somebody, surveillance of somebody by somebody, and fee collection by somebody)

....... The useful discussion is, identity and reputation infrastructure... that's the missing enzyme, enabling the world to move forward with the decentralization of its economic activities. In turn that depends on strong encryption. Webfunds client is a useful thought experiment in that it creates and manages its own keys, it teaches its students the general idea of having an identity, independent of any host, and using that identity to record symbols on other machines that are nonrepudiable, i.e. sufficently connected with the individual webfunds client to form the basis of a persistent longterm reputation.

Oh well. I must admit my understanding is only the result of a particular background, in accounting, then in the dotcom era when I was actually paid to learn and use UML, attend standards bodies like OMG, ebXML, for a few years,