Sounds interesting! I read about India's UPI for the first time a couple of months ago in an Economist article: https://archive.ph/WoqQp
Then the users and merchants can decide if they wait to click the verify button. And the default for anyone with data connectivity should probably be auto-verification. There is an incentive to verify, because if any double-spending has happened, the first to verify is the one who gets the money.
Double-spends will happen, because a user has to be allowed to log into a new device (if they lose their old one), and there is no way to know if the money spent from their old device was synced to the server yet.
You can solve for that by having short validity of the offline tokens on device and having that equal the cool-off period for reclaiming those funds when new device is provisioned. If I had ₹100 unspent tokens in my offline capable device that I lost, and got a new device provisioned the same day, then that ₹100 will show up in my account and be available to load on my device only after, say 7 days, of cooling-off period. In those 7 days, if someone who found your device could spend it and if so you lose it. If you know you are likely to be offline only sporadically for few hours and not for days, you can reduce the validity (and hence cooling-off period) significantly to just 12 or 24 hours.
Who would accept "provisional money" though, if there is a realistic risk of it being double spent and therefore effectively worthless?
> Double-spends will happen in any system that allows offline transactions
Only if you assume untrusted devices. That's why in most past and existing stored-value systems, smartcards are being used – these have different security properties.
It's up to the receiver whether they want to walk to the internet cafe and connect to verify it, or if they just trust the person they got it from.
If all the devices are trusted, but can be lost/destroyed and re-issued, you have the same problem.
But these aren't mandatory: Physical cash also can't be replaced when lost, and only rarely when physically destroyed. That model might be preferred in some scenarios.
Apart from classical robbery held at knifepoint or gunpoint and taken to an atm and being robbed or being blackmailed to actually do instant money transfer online which do happen, there are other lower threshold but more frequently occurring and more painful cash "thefts".
One eye-opening story I learnt a while ago was this: the women daily wage workers (or the household help that comes to my home) who are typically the breadwinners for the family are coerced by their drunk husbands to give their hard earned cash which they will waste on more alcohol or gambling. This was the situation for decades. Now, with zero-balance, zero-cost bank accounts being made available at scale in India, these women can now keep their money safe and provide care for their children by refusing to give money to their husbands (usually it involves telling them they don't have the money to give when in fact they do in their banks). Such is the social reality for so many.
I do not think the loss of privacy is worth the convenience when talking about replacing cash in a society.
Most existing stored-value systems use "shadow accounts" for reconciliation (in case one of the card-level private keys gets compromised and makes it possible to "double-spend" electronic cash), but that's not really a required part of the system. If it's secure enough, it's possible to just leave these out!
It's probably also possible to create some kind of hybrid system which does create pseudonymous traceable logs which offer some trade-off between privacy and security, similar to how banknotes have serial numbers which can theoretically be traced, but practically mostly aren't.
And the enormous advantage of a privacy-preserving e-cash/stored-value system over cash is that it works on the internet too.
See here for one example where the indian state denied citizens access to their money by freezing their UPI accounts after unproven allegations. https://www.thenewsminute.com/article/bank-acs-can-be-frozen...
With one or more bank accounts, the woman must now keep her phone safe and inaccessible from the husband too. Or she must be very quick and diligent in deleting all the SMSes sent by banks for all kinds of transactions. This adds more hassles than saving cash with someone else.
In all the cases of alcoholism and men snatching money from the women/household without any other care, it's violence that keeps that relationship and "contract" alive. That violence isn't going to disappear just because physical cash is replaced by a bank account.
In summary, I don't buy the conclusion you've quickly jumped to.
You can solve that by actually having the money in the device so that reconciliation is not required, digital cash like the abandoned Mondex stored value system.
In many existing systems, there's a split between "merchant" and "customer" cards, of which only merchant cards are allowed to accept payments. Merchants often have to go online once per day and batch-submit all of the day's payments in order to be guaranteed settlement.
Among other advantages, this allows for a "lost/stolen" card feature: If a user loses a card, that card can be denylisted on a global list synchronized to all merchant terminals daily. After another day of waiting for straggler transactions, it's possible to determine the remaining balance of the lost card from the backend, and reimburse its owner, since further funds on the card can no longer be spent anywhere in the system.
One day is essentially just an arbitrary timespan here – you could make it 14 days, or an hour. In the systems I've looked at closely, one day makes sense because these are historically transit-focused, and e.g. buses don't have continuous network connectivity, but do go to a depot in the evening, which is a good opportunity to clear and settle all of the day's payments.
Customer cards never need to be online in such a scheme.
lol, literally less than 20 seconds.
> Then, when does reconciliation happen? When either party goes online? What if it is not a closed loop system? Then, reconciliation needs to happen for both parties independently. How long can either party stay offline and continue to make transactions? Is it both send and receive or only send on one side and only receive on another side? What if one party (merchant/receiver) is more likely to be online (almost always). Does it just become online payments problem then? NFC tap and pay is exactly this scenario.
Again, literally the exact problem statement and value proposition of Lightning, but stupid, stupid me for daring to mention it here, I guess.
What conversation? You just dropped in here with flamewar bait and complaining about some self-imagined persecution without contributing any substance. Hence your comment is now gray.
Oh yeah, gee, that takes us right to it.
And fuck outta here with the self-persecution bullshit commentary. My comment mentioned crypto and received a downvote 20 seconds in and 3 within 5 minutes. This site is turning into a shithole and I'm not going to keep dancing around it.
thanks!
If both parties are offline, can you still make a secure transfer of stored value? Yes, with a trusted execution environment and cryptography, you can do it. So, yeah, smart card works; a smartphone works, a feature phone with NFC smart card works etc. For each, the risk profile is different which affects the answer to the next question.
Then, when does reconciliation happen? When either party goes online? What if it is not a closed loop system? Then, reconciliation needs to happen for both parties independently. How long can either party stay offline and continue to make transactions? Is it both send and receive or only send on one side and only receive on another side? What if one party (merchant/receiver) is more likely to be online (almost always). Does it just become online payments problem then? NFC tap and pay is exactly this scenario.
In India, right now, we have NETC stored value cards for closed loop systems like metro. For open systems like UPI we have recently introduced offline payments capability with very small stored value stored as tokens on the mobile phone and used for very small transactions. As the banks learn the real-world operational risks, the wallet limits and transaction limits will be increased.