Year-end issue (AI, Law, and Otter Things #20)
Hello, dear reader! Today's issue discusses otters, The Laundry Files (and environmental impacts of computing), and doctrinal legal research.
In the previous issue, I wrote a few words about the AI Act's treatment of "critical infrastructure". Right after I was done with those words, the Council of the European Union (do not mistake it for the European Council or the Council of Europe) circulated the Presidency's compromise text on the proposal. Among its many changes, the new text includes AI risks managing digital infrastructure into the list of high-risk applications in Annex III. So, it appears somebody in Brussels is reading this newsletter (just kidding!).
On a more serious note, some colleagues and I are launching a blog called The Digital Constitutionalist, which welcomes contributions on matters regarding constitutionalism in the digital age. We have three main sections — digital rights, platform governance, and the digital state —, as well as a section for essays and original work of science fiction that helps us think about issues covered by the three main sections. The blog's launch event will happen on 24 January 2022 at 5 pm (CET), and we will accept submissions shortly after that, on a rolling basis.
In otter news
Another piece of news related to this blog's editorial lines is that otters attacked a British man in Singapore. While walking in the Botanic Gardens, Graham George Spencer and a friend were chased by a pack of otters, and Mr Spencer was bit 26 times after failing to evade otters. According to the BBC, it seems that a runner had previously stepped on one of the otter cubs, and so the whole pack went after the nearest humans, who turned out to be Mr Spencer and his friend. It appears the victim received initial care in a Singaporean hospital and seems to be recovering well, so at least the worst outcomes have been averted.
This scenario reminds us of something I have previously stressed in this newsletter: as adorable as otters are, they are wild animals and not used to life in society. So, if you are ever so lucky to meet one of these adorable mustelids in person, remember to admire them from afar. Preferably, please do so in a place such as a wildlife sanctuary or a zoo, where we can be reasonably sure that they are being treated well.
The conditions, of course, depend on the species. There are many places for safely viewing a Neotropical river otter in Brazil, while Asian small-clawed otters seem to be very popular in zoos around the world (for my Tuscan readers, there is a couple of them in Pistoia). On the otter hand, sea otters seem to demand more in terms of logistics, being more common in installations such as the Lisbon Oceanarium or the large aquariums in the US West Coast.
To conclude this section, I will leave you with some otter cubs eating:
An environmental reading of CASE NIGHTMARE RAINBOW
Readers of this newsletter, especially those of a technological persuasion, might already be familiar with The Laundry Files series by Charlie Stross. First published in the early 2000s, these books offer a mixture of cosmic horror, singularity stories, and bureaucracy humour. The basic premise of the series is that magic exists in the modern universe. More specifically, magic is a form of computation: by performing certain mathematical operations, people may end up casting spells — or inadvertently summoning up a beast from another dimension. In this post, I want to discuss how a vital element of the series' narrative may help us think about the environmental impact of computing.
HIC SUNT SPOILERS
The idea that people can cast magic spells through computation lends itself to a good narrative. Programming can feel a bit like magic at times, even for insiders. But it also has some large-scale implications. Think of Moore's Law: if computing capabilities grow every two years or so, does this mean humans can cast increasingly powerful magic as time goes by? Stross, who is quite familiar with information and communication technologies, does not shy away from the implications of this point. Indeed, the later books of the series deal with the idea of a mystical Singularity: every computing operation weakens reality a little bit more than the previous ones, producing more powerful effects. In a world where ever-faster computer chips are present in even the most trivial appliances, this creates a vicious cycle that leads up to some World Altering Events.
Governmental response to computational magic takes a few forms, centred on two primary goals. On the one hand, government agencies seek to contain crises such as people finding out about magic or cultists trying to benefit from cloud computing. On the other hand, they need to direct resources to prevent existential risks, such as somebody summoning Cthulhu by accident or a misplaced for loop opening a gateway for invaders from another dimension. Personally, I am not really interested in the problem of existential risk — though there are very smart people working on related topics if you are into that kind of thing. Instead, I want to call attention to a specific point of the in-universe response to these crises.
Since (the fictional) governments are well aware of the risks associated with computing, their most extreme plans for responding to the crisis involve many restrictions to the use of specific technologies. Yet, these plans end up being implemented too late, as governments are pretty afraid of worsening their constituents' standards of life for a threat that would seem to be coming straight from fantasy. Eventually, computation-restricting measures are put into place, and, in light of certain geopolitical developments, they end up being downright tyrannical.
When I read about this, I could not help but think about our global responses to climate change, both because of the perils of eco-fascism and of the impact that a successful response to environmental challenges on social life. Though I agree with Branko Milanovic that degrowth is, at best, magical thinking, it seems unlikely that we will escape this situation without substantially transforming some sectors of our economy beyond the obvious targets (such as the fossil fuel industry). And computing is a sector that might be radically changed in this process.
Our current computing model requires lots of energy and a problematic supply chain involving precarious labour, environmentally hazardous mining, and many other externalities (don't get me started about blockchain). So it seems likely that computing-intensive activities will face higher pressure on this front. In fact, a joint statement by more than a hundred civil society organisations in Europe has pushed for the inclusion of environmental oversight into the AI Act. Measures of this are welcome sources of information that will allow us to better view the externalities of computing, providing information for future policy steps.
In this context, Stross's treatment of computation as a physical resource should not be perceived as an actual policy recommendation. After all, massive restrictions to computation are unlikely to be sustained without authoritarian measures. And other policy and technological factors, such as steering public and private actors towards more energy-efficient technologies, may contribute to a smoother transition of the digital infrastructure towards a sustainable model. Nevertheless, it provides us with a reminder that large-scale computation leaves equally huge footprints in the physical world, something that tech policy and regulation need to start dealing with.
My two cents about doctrinal legal research
It might come as little surprise that my experience as a legal researcher has drawn from disciplines beyond the law. Except for a brief flirtation with tax law at the beginning of my LLB and (surprisingly) an internship with a lawyer working on criminal matters, most of my law-related work drew from my background as a computer scientist. Still, I am the product of a civil law education, and so my image of legal scholarship ascribes a central role to doctrinal research.
Such an image, however, seems to be somewhat old-fashioned. In part, this impression comes from my current environment. The EUI has many researchers doing strong work in a broad range of methods — law and economics, socio-legal research, statistical analyses of law —, which lead to sophisticated views of law as a social phenomenon. Such approaches have long been accepted in English-speaking legal academia, but nowadays they represent a growing share of scholarship in civil law jurisdictions as well. In my view, this is a valuable evolution, but I am less sanguine about the accompanying movement of dismissing "merely" doctrinal research as limited or minor work that is better left to practitioners.
There are some interesting debates on whether legal scholars should prioritise research that supports legal practice with a top-down view of legal issues or, instead, focus on the kind of "deep" legal research that is unfeasible for lawyers and judges in light of their professional duties. I have my views on this subject, which I am happy to share over a few drinks but will only commit to writing in the not-so-close future, after I have the legal experience to lend gravitas to my perspective. For now, let's just say that I subscribe to a pluralist methodology and that division of labour can lead to a better portrait of the law than what we would get from claiming that there is only one road to legal scholarship.
A different line of argumentation, which I do not buy, is that the gap between law in the books and law in action renders black letter law irrelevant. If the gap is big enough to deprive legal text of even a shred of relevance, then it would make no sense at all to spend time with them. But even systems with poor or selective enforcement of legal rules — in democracies and sometimes in authoritarian systems — often try to couch arbitrary actions in the language of legalism as a legitimising move. In these cases, doctrinal thinking may have a symbolic value, both in restating a legal ideal that the practice does not live up to and depriving malicious actors of the cloak of legalism.
Black letter law by itself won't solve those problems, and no amount of legal interpretation will topple an authoritarian regime. Even in a democracy, it takes action from the civil society (see the Schrems cases) or within the government to effect change in practice. But the symbolic value of doctrinal studies may still provide a useful, if partial, contribution to understanding these phenomena and rallying action. And, in the cases where there is a smaller gap between practice and the books, doctrinal thinking is more relevant as both a normative and a descriptive tool. It still won't tell the whole story, of course, as the law is not impervious to factors beyond its internal logic (far from it). But, as a better thinker than I once said in another context, there is room for words on subjects other than last words.
This is it for today
According to our schedule, the next issue should be published on 29 December. But, since this is the end of the year, I will resume the newsletter in early January 2022, after a short winter break. Happy holidays!