Identifying banknote fingerprints can stop counterfeits on streets | Hacker News Reader

Identifying banknote fingerprints can stop counterfeits on streets

warwick.ac.uk

17 pointssizzle16 comments

So in the actual article they prove they can tell one real £10 note from another real £10 note, because every plastic note accidentally has a unique random texture.

But as these are real notes, they can be uniquely identified by the serial number anyway.

They never get to a part where they explain how this technique is useful in telling real notes from fake, since fake plastic notes would also have unique textures.

Associate the fingerprint with the ID in the “what money have we issued?” database, probably. This sounds like a more practical version of my “use leaves as money” idea, and that's what I did. The impressive part of this research is realising that there might be a pre-existing “fingerprint”, and finding a way of looking at it. They probably assumed “how this is useful” was obvious.

They would keep a DB of the note fingerprints and look them up in that. Thus, this would only work if online or something.

You can counterfeit that serial number. It's much harder, perhaps neigh impossible, to counterfeit this fingerprint. Thus a database of existing fingerprints tells you whether it's real or counterfeit. If you kept a database of serial numbers, you'd additionally need to store who has which bill or you'd still not know whether you're the sole owner of this serial number.

Couldn't the central bank just sign the fingerprint and print the signature on the note? If the fingerprint can't be counterfeited and the private key can't be stolen, then the note would be unforgeable. As long as the fingerprint is robust enough.

This was my thought, but with a central database of serial -> fingerprint. I like yours because of the distributed nature.

I assume they would fingerprint every note at manufacture, then compare every note at purchase time to see if it matches a known good fingerprint. You would need to completely remake all bills in circulation with a new style and make invalid the old style, and equip everyone that takes cash with a fingerprint machine. If they tried to just slowly scan in existing bills people would just keep making the fakes, I would imagine.

You do not have to equip everyone who takes cash with a scanner, as banknotes regularly circulate through banks.

I do not know how, and how frequently, current anti-counterfeiting measures are checked, but the more subtle of them are almost never checked in ordinary cash transactions. Presumably they are checked somewhere?

I would imagine the only time it is important to stop a counterfeit bill is the first time its used, right? If the goal is to stop people from counterfeiting, at least.

I don't think that has ever been realistic. Like credit-card fraud, check kiting and other petty crimes, it is only realistic to stop it growing to the point of being a widespread problem that threatens the economic infrastructure.

> You would need to completely remake all bills in circulation with a new style and make invalid the old style...

It takes a while for all the old bills to get out of circulation though -- a few years ago I got my hands on a $20 bill from 1970 and the kid at the corner store almost didn't take it since they had apparently never seen one that old, probably didn't help it was nice and crisp as well. Only thing I can imagine is someone was saving it from their birth year and had to spend it for whatever reason since $20 in 1970 was over $100 in today's dollars so they lost a bunch of value holding that bill.

I am not convinced that the texture fingerprint is robust enough to remain unchanged once the note goes out into the world and is circulated for a while. Notes would start to be rejected as not matching the factory fingerprint just from normal wear.

The article covers this concern, and the claim there is that they have verified that regular wear and tear does not tarnish the fingerprint.

Yes, they left out that part. The paper's abstract ends with:

>[Our method] ensures that even in the extreme case when counterfeiters have procured the same printing equipment and ink as used by a legitimate government, counterfeiting banknotes remains infeasible because of the difficulty to replicate a stochastic manufacturing process.

Which means each note manufacturer has its own signature, though whether the researchers confirmed that with more than random samples from circulation, I haven't been able to tell. The full paper is here: https://arxiv.org/abs/2103.06184

I was curious so checked and thought other might be interested.

https://www.bankofengland.co.uk/banknotes/counterfeit-bankno...

"How many counterfeit banknotes are in circulation?

The vast majority of counterfeits are discovered before they go back into circulation, when retailers and the banking system are sorting them. A smaller number are detected by the public or retailers who hand them directly to the police, or when the police carry out search warrants. Counterfeits are typically removed from circulation quickly, often after a single use.

Only a small fraction, typically less than 0.02% of banknotes are counterfeit, that is less than 1 in 5,000 banknotes. In the first half of 2020, around 94,000 counterfeit Bank of England banknotes with a face value of £2.1 million were taken out of circulation. At any one time, there is an average of 4 billion genuine banknotes in circulation, with a notional face value of around £75 billion."

The scope of the problem seems to be about 2 million pounds per year. I wonder what this solution will save and cost.

I have read numerous times that card sharps from days gone by became proficient at noticing the various signature discrepancies of the texture and inking on the back of playing cards so they would not need to crimp, clip or ink a card to mark at least some of them in a particular deck in play.

Basically the more proficient you were at spotting these details, the more "marked" a deck might as well be for you.

Reply on news.ycombinator.com