Name:
Blockchain
Description:
Blockchain
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/cec06b61-2448-4eff-ae9c-1409a317e7d6/videoscrubberimages/Scrubber_10.jpg
Duration:
T00H12M45S
Embed URL:
https://stream.cadmore.media/player/cec06b61-2448-4eff-ae9c-1409a317e7d6
Content URL:
https://asa1cadmoremedia.blob.core.windows.net/asset-51871322-f754-4019-a626-77002dd09164/Track 1 - Ian Mulvany_1.mp4
Upload Date:
2022-05-02T00:00:00.0000000
Transcript:
Language: EN.
Segment:0 .
IAN MULVANY: Hey, everybody. I'm Ian. I work at SAGE. I'm head of transformation. So what I do at SAGE broadly is trying to understand how technologies and modern development methodologies can help us in developing new products or develop products efficiently or transform the way we're doing internal processes. And I have a background in a number of innovations in product development in this industry and in the sector.
IAN MULVANY: So I'm going to talk about a blockchain. My key message is just run away from it. Don't use it. It's awful. Yeah. And we could just stop now, and I could save you some time. But I'm going to go into the depths of why I think it's completely unfit for our industry. So this is the takeaway.
IAN MULVANY: While discussing interoperable metadata and interchangeable standards is a good thing that blockchain might make us do, it is inappropriate as a technology for industry because it is too complex. It requires lack of trust-- we're a very trusting industry-- to be valuable, and it does not offer as much potential ROI as almost any other kind of technology that you're going to stumble across when you walk out your front door.
IAN MULVANY: But why are people talking about this? So I've just pulled two quotes with two different perspectives from Phil Davis. He's going to propose this as a potential solution to the growing problem of authentication and accounting in scholarly publishing, and it may be found in bitcoin or blockchain. Whereas Manuel Martin says that, "Third-party companies have made huge profits from the work of scientists for many, many years selling journals and reprints." Blockchain will make it possible for authors to take back control.
IAN MULVANY: So basically, one person thinks blockchain is going to kill sci-hub. Another person thinks blockchain is going to kill publishers. And I will put it to you that both are unlikely to be true. All right, so this used to say Blockchain-- A One-Minute Primer, but it's taking me a bit more than one minute to go through. This my x minute primer on blockchain, where x is less than 10, hopefully.
IAN MULVANY: So I'm going to explain blockchain, and then I'm going to explain the fundamental things you need to operate a blockchain. And there are four of them. And I think each four of those are inappropriate for our industry, and then that'll be a wrap. OK. OK, so who here knows what a blockchain-- who feels confident that you know what a blockchain is?
IAN MULVANY: OK. So we have some people. So if I get it wrong, please let me know. So what we start with is we start with a packet of metadata and data. It could be a paper. It could be record in your database. Doesn't matter. And you create a cryptographic hash of that.
IAN MULVANY: And so this is a small representation of this. It's a string. It's unique. If this changes, that changes. And then we want to add our next element. This is our founding component of our blockchain. So our next element is, again, we have the next paper, for example, or maybe the next peer review record.
IAN MULVANY: But when we create the hash of this, we include the hash of the previous item in this chain. And then we go on, and we include the hash of these things and this object as we add it to this chain. And so when we talk about a blockchain, we're talking about blocks of data that are chained together through these cryptographic hashes of the previous elements of that chain.
IAN MULVANY: And then, you know, you just you just keep going. And so one of the really interesting things about blockchains is they're immutable. And in them, they contain the record of every single transaction that has happened up until that point in the history of the blockchain. So unlike your database or your SQL database, or even a publishing record, which might have tombstoning or retractions, the blockchain is supposed to be immutable over time.
IAN MULVANY: It contains the entire history of what went before it. And then we create a network of nodes who are this, like, distributed authoritative list network. And each of them keeps a copy of that blockchain. And they come to some method of agreement around what is the source of truth, or what is truth in that chain? Now, let's say I'm some sort of evil actor, and I'm like, you know, I want to change the record, and this might represent transactions of payment or moneys, or I might say, you know, I'm going to say that here, I've actually got more citations than I really got, and I'm going to go back into history, and I'm going to change the record.
IAN MULVANY: Well, because this cryptographic hash changes when they change the content here, and all of these are stacked on top of that one, everything else in this chain onwards is no longer going to match the other copies in the network. It's going to be different. OK. But what happens if I'm really evil, and I'm able to access all of the nodes and change all of those copies at the same time?
IAN MULVANY: Well, I'll get to that in a moment. But when new data comes onto the network, we have to have an algorithm for how the network itself can choose which of the new versions of record it agrees on. And what happens is they all ask their nearest neighbors to agree with them. Data starts to flow through the network, and whoever manages to capture 51% of the network, the algorithm on how you run a blockchain says, once you get to 51%, everyone flips to that voting.
IAN MULVANY: And so it's sort of majority voting. So if I were evil, and I were able to attack 51% of the nodes, I could take control of the network of the information on that network. And in a distributed system, where we can actually have signals that are transferring really quickly, that might potentially be possible. And so this is another piece of the blockchain puzzle, which is transactions in the network are designed to be slow in order to prevent someone from being able to attack more than a few nodes simultaneously.
IAN MULVANY: And this is the thing called proof of work. And you may have read about how blockchain or bitcoin is using an enormous amount of energy. It's using like more energy than the annual energy use of Ireland to support the transactions in a year. And the reason for this is a dumb, pointless, useless, numerical challenge is posed before any node can take a new piece of information on it. And this is the proof of work, and it is purely designed to stop people being able to attack the whole network quickly.
IAN MULVANY: So it's this slow system for transactions. So what we need is we need four things to create a blockchain network. We need distributed writers, writing lots of copies of our entire database. We need a network of nodes who have multiple copies of those things. We need an algorithm of proof of work to prevent 51% attacks. Because all of this is actually fairly computationally intensive, we need some kind of rewards incentive system for people to participate in this.
IAN MULVANY: Otherwise no one is going to try. And I think in scholarly publishing, that we don't have those things. I think that it is fairly hard for our organizations to get to consensus on anything. We have good solid metadata rules. So just go look at the state of metadata 2020, which we should all participate in, who have been trying to get the metadata of our organizations to align for years.
IAN MULVANY: We're still not there. It's a good thing. We should all try and do that. But we haven't got there yet. How are we all going to get agreement on running these nodes simultaneously? We actually are a high trusting industry. We are working in a high trust environment. We don't need to build on technologies who are designed for situations where there is low trust, and there are plenty of situations which are low trust, where this technology is more appropriate.
IAN MULVANY: But ours is not one. The kinds of attacks that you might apply to the scholarly record, I think, are fairly pointless, because we don't tend to have transactions that are anonymous. People tend to stand by their scholarly record, and when they get found out that they've been doing something bad, this is uncovered, they get debarred.
IAN MULVANY: And so there's already low incentive for an attack on this record. So you don't need to have a prevention against such an attack. And then finally, this is the one that really gets me. I've read so much about using blockchains in our industry and how you can generate coins and these tokens would represent like peer review credits or this credit or that credit.
IAN MULVANY: We have an incredibly entrenched system for incentives already. Many of you have attempted many things to try and move that in any way, shape, or form. It is really hard to move the incentive system in our industry. I am not convinced that X-coin or Litecoin or PSIcoin will do that. I could be wrong. But this is where I stand on this position at this point in time.
IAN MULVANY: All right, so the idea is fairly simple. But it's actually really hard to implement cleanly and neatly without risk. And so I'll just give a case study of bitcoin, which I'm sure most people have heard of. There's this idea that network nodes, anyone can participate, and therefore, it is an egalitarian system. But in the case of bitcoin, it's possible that 95% of the transactions on the bitcoin network are actually fraudulent and are not from regular players.
IAN MULVANY: And so this goal of egalitarianists does not hold true in bitcoin. There's this idea that blockchains will efficiently and evenly distribute the load, and anyone can participate equally. But 80% of the transactions in the bitcoin network is performed by one of six mining pools, five of which are in China. And so it's not actually a decentralized system, even though it purports to be.
IAN MULVANY: And then there's this sense that blockchains should be secure and tamper proof. And yet, in 2018, $2.7 million were stolen out of different caches in the bitcoin network. And the reasons for this thing at the bottom is that the algorithms, although conceptually easy, are actually extremely difficult to program robustly. OK. So the risks are that you'll get hacked, and the other risk-- and I think the second is actually the biggest one in our industry-- you all say, well, let's try this project with bitcoin.
IAN MULVANY: Please try projects with their AI. AI's fantastic. You try project with blockchain, there are a small number of vendors, and we'll end up with the same vendors providing the same solution for all of our different organizations, but that's not distributed. That's vendor lock-in. We have a lack of technical capacity within our organizations to understand how to run these things.
IAN MULVANY: And then what is the transaction that is in there? Who owns it? Who owns the peer review? We don't have answers to the fundamental questions of the things that would even exist on these networks. I've already mentioned my concerns about reward systems. Now, I think there are some benefits, because you're supposed to think of things that are potentially good about these things when you give talks like this.
IAN MULVANY: So I came up with a list of things that could be good. So one thing I have observed is it has begun conversations amongst many stakeholders to talk about what are good representations of metadata of some of the more complex kinds of transactions in our space. And I think that's a really good thing. We should be talking about metadata. I really like the idea of creating a store of events that have happened and thinking about the scholarly record as it append only store things that have happened.
IAN MULVANY: And I think this sort of a sense of a global append only store has a lot of potential really interesting things that you could build on top of. But I don't think you need blockchain to be able to do that. I like the idea of submissions as transaction events, multiple copies by design. These are things that I think are good. But in terms of technologies, you could get to the same thing using any number of other really interesting technologies that I won't discuss now.
IAN MULVANY: And if you were looking for things beyond blockchain, I would recommend spending your money on metadata standards, on other kinds of distributed systems, on open publishing systems, and on machine learning and artificial intelligence. And I'll just leave you with this slide, someone saying, "Bitcoin is money for people who believe in math and software." And a software engineer replies, "As someone who writes software for a living, I cannot exclaim strongly enough just how little I believe in software." And so that is-- that's what I want to say about blockchain.
IAN MULVANY: Thank you. [APPLAUSE]