Fake Whisky
How do you guarantee provenance? Why are evidence trails difficult?
For more information on Aleph Insights visit our website https://alephinsights.com or to get in touch about our podcast email podcast@alephinsights.com
Transcript
Hello, and welcome to the Cognitive Engineering Podcast produced by Tell Me Studios for Aleph Insights. In this series of podcasts, we take a look at interesting topics and discuss what we think they tell us about analysis and decision-making. I'm Fraser McGruer, and I'm here with Chris Wragg, Peter Coghill, and Nick Hare of Aleph Insights. And this week, we're discussing fake whiskey.
Speaker B: was supposed to be a Macallan: Speaker A:Okay, before handing it over onto others, though, did you come up with any initial answers to your questions about provenance?
Speaker B:Well, computers and things, so if you're being investigated and there are a number of MPs being investigated for various reasons in Parliament again, they're often particularly interested in transactional data, so who sent what messages to who and when they sent them and things like that, because with that kind of information, you can build up an accurate picture of who was doing what when, and in what order those things occurred. Now, you can set up computers to maintain a log, so say log in, it will log who logged in when, but then those logs could also be tampered with, so you'd need another log to log when the logs have been changed, so it's incredibly difficult, so the only way really to do it was the way computers are currently mostly configured is you have to seize it, you have to take it out of the hands of somebody who wants to have any reason to change it and impound it so that it can't be accessed, which is difficult because there's plenty of times when the accusation might be fairly minor and you need that computer to continue doing your job, so how do you seize those logs? And it becomes even more challenging when if the data that's important to the investigation is distributed across lots of different servers in the cloud, how do you seize that data and take a snapshot of it so that it can't be tampered with?
Speaker C:Yeah, I mean, for me, I think this issue of provenance is really interesting and it's obviously very difficult to do and I think one of the key questions is not how do we do it, but when do we need to do it, so in the case of the whiskey, for example, did it actually matter that it wasn't the vintage that he said it was, if the chap had enjoyed it and drunk it, there are all sorts of examples where actually for something like that that's for our pleasure, the provenance doesn't matter and in fact, being told the provenance is something that it isn't actually enhances the experience. So with taste, the newly emerging field of neuro-gastronomy is actually starting to realise that taste is not something that's inherent in the food, it doesn't exist within the food, it's something that's created by our brain based on the sort of sensory input we get, but also the preloading we get in terms of being told about something.
Speaker D:Well, I mean, come on, I mean, you're not gonna convince me that Marmite on toast is a glass of wine, I mean, that's not... No, but you can be... It's obviously strongly influenced by the thing you're...
Speaker C:Yes, by the visual picture, but there are examples, for example, where people have been given chocolate yoghurt in the dark and told to taste test this new strawberry yoghurt and a sizeable proportion of the respondents have said, oh yes, I particularly like this, the really strong strawberry flavour of it or examples where experts have been, wine experts have been given white wine that has been dyed to look red and simply not spotted that fact, let alone some of the more distinguished things. So no, if you've got visual reference, you've got smoke, it's a whole sensory creation. If you've got all of that and you're being told something, you probably could, in some circumstances, convince somebody that something that tastes strongly in one way actually tastes another way.
Speaker D:Sure, no, but I think that we shouldn't, I mean, the fact that we can show that our perception of taste is influenced, I mean, it's no different, really, the perception of colour, for example, you remember the dress from a couple of years ago, was it blue or black or was it, you know, and... Or beauty, what's beauty? Well, but I mean, these are things that are a bit more objective. I mean, you can use different kinds of shading, for example, to make two things that are exactly the same colour look like completely different colours. And yeah, it's influenced by context and all sorts of other things. But it doesn't mean there's not fundamentally some underlying physical cause here. But the question is here is, you know, fine, so that's the case. It seems to me perfectly okay to want to pay lots of money to taste a whiskey that is historically significant. It's like, you know, imagine drinking brandy that was owned by Napoleon or something. Now, that will be interesting. I don't want to pay to believe that it was owned by Napoleon. I want to pay because it was actually owned by Napoleon and I will enjoy it more knowing it's owned by Napoleon than, you know, if I don't know that.
Speaker C: nking it, he knew that it was: Speaker D:No, no, but in that context. Yeah, but in terms of his experience, like his sensory experience might have been the same. But this is like saying, you know, it's sort of, you know, would we be happy to be plugged into a sort of matrix type thing and where we merely believed we were in the real world eating a steak? Or do we assign some value to actually experiencing what we think we're experiencing? And I think we do. I don't think we, we don't just want the experiences of certain states of the world. We want those states of the world to pertain. And with you, you might think that's stupid, but the fact is it's true that people do have preferences about the way the real world actually is that aren't to do with preferences about what news they're receiving about the world. Right? I mean, obviously.
Speaker B:If they're sensory, if the stimulation is absolutely equivalent, then how would you know? How would you know that?
Speaker D:No, but that's a different question. It's how I'd know is what we're talking about. It's different from saying, well, they're the same, right? Believing wrongly that a whiskey is 150 years old is not the same as actually knowing it's 150 years old because it is. Now, how you tell the difference is, I think, this question of how we establish provenance, but they're not the same experience. And even though they might feel exactly the same, they might be indistinguishable, there is obviously a difference. People have preferences for what, for it being true. I mean, the question of how they establish it's true is a different thing.
Speaker C:Nobody wants to be deceived. Right. But in the event that we are deceived and we never find out about it, it's indistinguishable from the other.
Speaker D:Yeah.
Speaker C:But I think coming back to sort of the issue of provenance and when or why we establish it, there are obviously cases where it's more important to establish provenance than otherwise. So for example, with currencies, when we make a financial transaction, we want to know that that financial transaction actually relates to some real money somewhere that's going to turn up in our account and that we're going to be able to use in the future. But there are other areas where, if we're having a conversation with somebody online or something and they're lying about their identity and they're actually a bot or somebody, some Russian troll posing as somebody, where it may be less important in a trivial conversational context, but in a, I'm going to now influence the outcome of this election context, it may be more important. So it's, you know, provenance is obviously difficult. And, you know, Peter talked about the difficulties of doing it in a digital environment. And because it's difficult, it probably makes sense to think about, you know, when we definitely need to know the provenance of something and when we can let it slide a bit.
Speaker D:Yeah, I think that, you know, there's a basic kind of theory of provenance, which doesn't really treat it as any different to any other kind of information problem. What you're trying to do is find some signature, some aspect or feature, if you like, of an object or a transaction or any other thing that is very improbable to have come from something else. You know, it's very unlikely to see that signature in a non-authentic article. And highly likely to see it in an authentic article. That's essentially what you're trying to do. And usually what you find in these sorts of cases is that it's quite easy to rule stuff out generally. So actually in the case of this whiskey, as far as I'm aware, even the label was wrong. The label wasn't the actual label McCallum were using at the time. So that should have been very easy to rule out. But often it's quite hard. These things are quite hard to sort of rule in, to be absolutely certain about. And I don't know if anyone here watches Fake or Fortune, but it's a very interesting program in which the BBC investigate whether or not a particular painting is genuinely by say Monet or someone. And of course that has enormous impact on its value. And usually it's quite, it's a sort of, they end up at this frustrating situation where, well, nothing's been ruled out. So you have the labels are right, the paint is right, the style is right. But there's just, it's nothing actually saying that Monet painted this somewhere. You know, it's normally where they get into that sort of zone. And sorry, yes, Fraser.
Speaker A:Well, I just wanted to bring it back round because we were talking, I know we led in with the whiskey and actually that's a bit of an outlier because it should have been pretty easy and pretty swiftly to work out that it's provenance. But just going back to this digital question, you know, is there anything that's different now than might've been 10, 20 years ago in thinking about digital information? And I mean, what can we add to the discussion on that? And is it more difficult or easier to prove provenance or to define provenance now?
Speaker D:So I think there is, I think Peter's right in the sense, you know, that if you think about what we're at, what information stored in electronic form looks like, it's very mutable. You know, it's essentially switches, tiny switches being one way or the other, and you can just come in, switch them into a different configuration. And there is almost nothing to tell you that they used to be in a different configuration. So at a fundamental level, you can't do that with a book. You know, you can't take a new piece of paper and do things to it, you know, easily that will make it look like an old piece of paper. Just, it's not, you know, that's just, it's physically very challenging. Physical objects are very hard to change on a fundamental level. Whereas information as stored in an electronic form is easy to change. So what do we do about that?
Speaker B:Peter, what do we do about that? Well, it gives us a chance to talk about blockchain again, because the blockchain presents us with possibilities that we haven't had hitherto. A distributed ledger that records all or any transactions could be an answer to this. So, because you're not relying on a central authority, which if vulnerable could mean the doctoring of records, of logs, because everyone potentially has a copy of all the transactions, then it's much harder to change in a sort of analog to the difficulty of changing a physical object.
Speaker D:Yeah, I think this, what it's essentially doing is it's like it's sending out, you're churning out a lot more information every time and throwing it around the place every time you want to record a transaction. And in a sense, you've got a lot more bits of thing to hunt down. You've got a lot more things to hunt down and change if you wanted to somehow defraud the world through a blockchain. It's certainly not totally impossible to imagine some software that could do that, that could subvert a blockchain. It's just very, very hard. It just makes it harder because there's more of it. And I think that's the equivalent to, there's no reason a centralized authority couldn't do it, but this is the equivalent of taking lots of copies and sending them to people. You know, it means that it's just, there's a lot more work you've got to do to try and commit fraud.
Speaker A:I'm not sure if we're getting close to any kind of resolution on this, but we are close to needing to finish. Chris, anything to weigh in with at this stage?
Speaker C:Well, I think, you know, at the heart of something like blockchain, yes, it's about electronic, you know, records and the use of a distributed ledger, but it's kind of at its heart is still, you know, the verification of trust in something by a distributed network of people. I mean, people are attached to those computers, but it's kind of the same as having a mass, a mass witnessing of something, you know, that it's much more difficult. You know, you might go and change one person's mind, but you can't change, you know, millions of people's minds that they didn't witness something. So I think, you know, it's kind of taking an extension of that, of the mass witnessing of something and putting it in a kind of digital format. So I think it's a technology that works with people and people's ownership of information. So I think that's why it's, you know, potentially going to be more trustworthy than other technologies.
Speaker B:Yeah, and it relies on a mass of people who have copies of all of these transactions, who are indifferent to specific transactions, but are interested in the security of all of the transactions collectively. So they will take steps to make sure that their system can't be doctored because they have a general interest because the integrity of the whole blockchain is important, but they are indifferent about specific transactions represented there.
Speaker D:Yeah, I mean, so, I mean, this, which touched on the reason why we care about provenance in some cases, and it's almost always the, we need to have a provenance technology when there is an incentive to fake provenance in some way. You know, whereas things like, you know, you look at something like trying to authenticate some, you know, whether or not some artifacts are Roman or not, is something no one probably really has that big an incentive to try and fake, right? So generally, there's, we don't have to develop a technology which is hard, which is unbreakable because people generally don't have an incentive to fake those kinds of things. But obviously things like art or whiskey, where having a good provenance, it makes something hugely more valuable, then there is a competitive element to it, which means that's where we need to focus our kind of.
Speaker A:Yeah, or just buy off the shelf whiskey.
Speaker D:Which will probably be just as good. Yeah. I think in most cases.
Speaker B:But if you're interested in investing in whiskey, then that's not. Investing is a different matter.
Speaker D:As far as I'm aware, the advice is don't, if you're trying to get a return on it.
Speaker A:Okay, we need to finish there. Chris, do you want to come in with something?
Speaker C:I was just going to sort of conclude from my perspective with the quote from Theodore Rousseau, the famous art historian and forgery connoisseur. Okay, excellent. Who said that we only find out about the bad forgeries, the good ones are still hanging on the walls, which I think is worth us bearing in mind from a humility angle.
Speaker A:Very apt, very apt, very apt. Nice note to finish on. Okay, well, thank you as always for listening to the Cognitive Engineering Podcast. I'm Fraser McGruer. We've been here with Chris Wragg, Peter Coghill and Nick Hare of Aleph Insights. Thank you. And until next time, goodbye.
