Future Tech? Yes please

People keep talking about big data and how important it is. They talk about how useful it is in making predictions and infographics, or how much it can be worth. But how does big data affect individuals like you and me?

Big Data: Data sets containing much larger amounts of data than what would normally be processed. It can be in the terabytes or up to zettabytes of information gathered from many sources and about many things. It is usually defined by the volume, velocity and variety of the data.

Big data has many uses. It can be used by companies like Netflix to anticipate trends based on user activity. They look at things like how people have acted in the past, what affected how they acted and what is happening now. It's also used to find flaws in things by seeing where people struggle with them and where they tend to fail. Big data is used to train Artificial Intelligence (AI), and to maximise internal operational efficiency. According to the UN Environment Programme, big data can help assess environmental risks and minimise the harm done to it.

According to IBM, Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind. That means that essentially all it has to be is an algorithm that makes decisions to qualify as AI. AI can then be "weak" or "strong", where weak ones have a single or narrow purpose. For an example, consider Deep Blue, the chess computer.

Like every other seemingly perfect solution, big data isn't without its risks. First we have to as ourselves how exactly they are getting this big data. Such huge amounts of information have to come from somewhere, and that place is usually from us. Most things you do can be and often are tracked. If you go to websites, not only will they, Google and often Facebook see that you are there and everything you do there, but often they will have additional cookies track you across the internet afterwards. If you use the Wi-Fi at a café you visit, or ask Alexa about the weather in Paris, that may be saved. These reams of data can then be used to try to answer any questions companies may have about you, like how you would respond to specific adverts or if you are likely to buy something.

The General Data Protection Regulation (GDPR), a fairly recent EU regulation that controls data use, has as some of its core principles Data Minimisation and Purpose Limitation. The former is the idea that companies should only have the information they absolutely need to do their job. It clashes quite directly with big data, which necessitates having as much data as possible. The latter also clashes because it is to only use data for a set, closed purpose whereas big data gathers the data first and uses it to answer questions later.

Things get even more complicated when you start thinking about the kind of AI that learns on its own. They use big data to change themselves, and then make more decisions based on big data. This makes them minimally transparent, and increases the risk of bias. Especially if the data it is learning from is skewed, its predictions may quickly become discriminatory. For example, an algorithm trained on the massively Black and Latin NYPD gang ties database will almost certainly flag up far more minorities than white people in the future. Also, an algorithm being used in healthcare to sort people by how much help they needed gave equally sick black people lower risk scores because of the specific terms it used.

Can the AI making decisions based on big data ever really be accountable either? Accountability is another GDPR principle, but it becomes complicated when the processing is automatic and at such a massive scale. Who exactly is responsible for the results of the processing? Who is responsible for how the data was acquired? How do people know what data is being processed and what it is being processed for? The ICO also says that for an AI's processing to be accountable, bias has to be avoided before it affects people rather than spotted after. How do we do that?

The ICO is the Information Commissioner's Office, the entity whose role is to ensure data protection and that the GDPR is followed. They are the ones you can complain to if you feel that your data rights are being infringed.

All in all, while big data has a lot of potential, it also has many risks. If we can ensure that the data does not relate to humans when it does not have to (such as when it's about the environment) and that when it does the specifics are clear to every subject and they have given their consent then big data has the potential to make the world a better, more efficient place. However, we need to be very careful that the price of efficiency is not our data safety.

This is an excellent video on how companies track you. Most things you do online and increasing amounts of the things you do in person are fed into databases that are then used to tailor adverts to you and when banks or potential employers make decisions about you. One of the most common online tracking methods is through cookies, which you most likely accept whenever you enter a website.

The advent of Bitcoin and other cryptocurrencies has brought the concept of blockchains to the attention of the public. You'll hear it in the news, at corporate events and whenever anyone is talking about innovation or technology. Some say it is the future, others say it is overhyped. Personally, I am not convinced that Bitcoin is all it is hyped up to be but the potential of blockchains is undeniable. Today I am going to discuss one of the potential uses of blockchains: smart contracts.

What is a blockchain? Originally, people used ledgers to keep track of transactions. The issue was that they could be changed or lost and only a few people had copies. A blockchain works by having copies of each transaction in identical chains saved across a network of computers. If one goes down or changes a block, the others can show the truth. Every so often, all the new transactions are verified and saved as a new block. Because the blocks have a part of the previous one in them, you could also not just remove one in the middle.

Smart contracts are essentially contracts written into a block in the blockchain. They are written as if-then statements, so for example if you transfer 20,000 GBP into a specific bank account the title of a car is transferred to you. The difference between this and a normal contract is that it's instantly, automatically and irreversibly executed when the condition is fulfilled.

Smart contracts have many potential benefits. They are instant, and importantly in cases with a large volume of contracts, they do not require any paperwork and avoid human error. The fact that they are on a blockchain means they are quite safe because the content is encrypted and to change anything you would have to control more than half the connected computers. They also avoid needing third parties, which lowers both the risk and costs. Finally, the transactions are fully transparent to both parties involved.

However, I suspect they will not replace all contractual transactions. In fact, I would say they will only really be used for routine sale agreements and small scale transactions. One reason for this is that most commercial contracts are significantly more complex than "you send me X, then I will send you Y". These contracts can be hundreds of pages long, so converting them to an instantly activated bit of code seems not only unlikely but potentially undesirable.

To add to the undesirability for specific types of contracts, the fact that smart contracts are so transparent and objective and instant can mean they lack a human dimension to them. There is no space for any leeway, renegotiation or interpretation, which is not always a good thing. If your contracts are complicated this could be an issue, and also if the contract is between, say, you and a family member, where you don't necessarily want it to be extremely strict.

Furthermore, because the smart contracts are simply chunks of code, they rarely have much interaction with the physical world. They might be able to check if money has left and/or entered a certain account, and trigger the transfer of a title or tell people to start working on something. What they struggle much more to do is check if complex, real life conditions have been fulfilled or keep track of subjective things like reputation. For this they rely on oracles, but that reintroduces human error and a third party into the mix.

What is an oracle? Oracles are companies that verify if conditions have been fulfilled, and then inform the smart contract if they have. For example, someone could hire an oracle to inform a smart contract when they die, triggering it and allowing it to automatically share the content of the will.

Finally, because both blockchain and smart contracts are in the early stages, there are many potential social and economic side effects and likely more that we have not even thought of. One that smart contracts may have is that if they are used for digital rights management (see my first post for more on that) they could allow companies to track all their content anywhere on the internet which might limit its open nature.

As it stands, companies like Facebook are not liable for content that is posted on their site. They may have to take down content they're aware of (such as if you report it to them) but under the eCommerce Directive Article 15 they are not obligated to proactively monitor uploaded content. This is justified by arguments such as that it would be impossible for YouTube to monitor all of the 500 hours of content that gets uploaded to it every minute.

The issue is that part of that content could be child abuse material, copyrighted content or hate speech for example. The easiest way to stop this would be to have the social media companies have to remove it/stop it from being uploaded or face sanctions. But how are they meant to go through everything? Can they afford to without crumbling or charging users? And finally, if they have to pay every time they let illegal content through, isn't there a good chance that they'll remove a lot of legal content rather than take the risk of not removing something illegal, thereby limiting freedom of expression?

For those who don't know, Digital Rights Management (DRM) refers to technologies used to block or limit the sharing or use of forms of digital media such as games, music or ebooks. They are a response to the challenges posed by the development of technology and how much easier it has become to pirate content.

In 2012 the Court of Justice of the European Union (ECJ) decided the case of UsedSoft v Oracle. UsedSoft sold second had software licences, such as the one by Oracle. Oracle's software licence allowed for 25 users, so UsedSoft would sell access to the software for any left over spaces. UsedSoft argued that the licence was paid for so exhaustion applied.

Exhaustion is a legal term meaning that when you sell something to someone you no longer control it and who they can sell it on to. For example, if you buy a physical book from the publisher they cannot stop you from selling it second hand.

The question of exhaustion was much debated, in part because there were two statutes that could apply. Both the InfoSoc Directive and the Software Directive say that the right holder holds the exclusive right to distribute the work (that is, the creator or original owner) but both also say in Art.4 that that right is exhausted after the first sale. The ECJ decided that this was fulfilled when it comes to software if the acquirer has paid the fair amount for an indefinite (or effectively indefinite, like 99 years) licence.

So far it seemed all was good. We could have a second hand market for software, allowing for example for people to sell on games they had beaten. However, and this is where it gets complicated, the court decided in 2018 that this only applied to software and not to other forms of digital media such as ebooks or music. In the case of Tom Kabinet, the eponymous company would allow the members of its "book club" to download any of the ebooks it owned. They were supposed to not make copies and to delete it when they finished and pass it on, but this was not effectively policed.

The court decided that rather than falling under art 4 (distribution to the public), it fell under art 3 (communication to the public) This is significant, as the Directive is clear that while the copyright owner's distribution right is exhausted with respect of a copy of the work by the first transfer of ownership of that copy with the right-holder's consent, so that the purchaser is free to resell it, the right of communication to the public is not subject to exhaustion.

The ECJ differentiated this from UsedSoft by saying that sending a program is the same as buying the CD original for one, whereas sending an ebook is different to getting a physical book. They decided that software falls under the lex specialis of the Software Directive, but digital media does not because it is not software.

Lex Specialis refers to the rule in law (mostly used in international law) that a more specific law trumps a more general one. If there is a law to do with food production and it clashes with one to do with the cleaning of beef, the more specific latter one would apply.

This raises many problems which have yet to be resolved. Firstly, the ECJ said that receiving a physical book is different to receiving an ebook. While that may be true, what about other forms of digital media (Angelopoulos, C.L.R. 2020)? Buying an album of Nickelback in CD is effectively the same as buying the album online and downloading it, especially nowadays when music is rarely on CDs.

They say programs should be treated differently to other forms of media because they are harder to share, however anyone moderately dedicated could copy a video game on their computer (Van der Donk, E.I.P.R. 2020).

They also say that you cannot sell intellectual property other than software (such as music or books) second hand, but seem to forget that it is almost impossible to find software that does not include IP (Ross, Ent. L. R 2020). Can I sell someone my second hand game but not the music on it?

Why does non-software intellectual property deserve more protection? Why should the people making video games have to accept the loss of income from having people buy second hand instead of from them but authors shouldn't? You could say that it's because there is lex specialis, but the question here is if the law is right, not what the law is, and if you did say that you are a boring person.

Intellectual property refers to the non physical property protected by patent, trademark, copyright etc. Copyright for example protects books, music, plays or software; and patents protect inventions. They make it so people cannot for example write an exact copy of A Song Of Ice And Fire and sell it.

Furthermore, in Tom Kabinet there was the issue that the company did not really make sure that only one person could read the book at a time, and that they deleted it before passing it on. Angelopoulos believes that if the ebook sharing was done properly, which is absolutely possible, the ECJ would likely allow it. It would be much more similar to how once you sell a book second hand you cannot read it again, and the ECJ seems to like that (see Vereniging Openbare Bibliotheken C-174/15).

So really it’s all fairly unclear, but we’ll see how things change and if the ECJ ever decided to make sense. Until then, it seems like most companies are going to be quite careful with any second hand sales, but maybe some enterprising UsedSoft 2 will come along.