11 January 2023. Tech | Innovation (2)
11 January 2023. Tech | Innovation (2)
The beginning of the end for personalised advertising // Surviving in wicked learning environments (more on ‘Range’)
Welcome to Just Two Things, which I try to publish three days a week. Some links may also appear on my blog from time to time. Links to the main articles are in cross-heads as well as the story. A reminder that if you don’t see Just Two Things in your inbox, it might have been routed to your spam filter. Comments are open.
1: The beginning of the end for personalised advertising
There’s a piece at Wired magazine by Morgan Meaker on a recent EU ruling against Meta that comes with a large fine and an instruction to the company to change its rules on personalised advertising on Facebook and Instagram. The fine is substantial (€390 million), at least by other people’s standards, but the legal ruling is far more damaging to Meta’s online advertising model.
(Jasper Johns, ‘Target with four faces’, 1955. Photo: Steven Zucker/flickr, CC BY-NC-SA 2.0)
Up to now, users have agreed to accept personalised advertising when they also agree to Meta’s lengthy Terms of Service—without which you can’t get access to Facebook or Instagram. Effectively, Ireland’s data watchdog—which oversees Meta’s European operations because they are based in Dublin—has now ruled that this is a breach of the EU’s GDPR rules, and that agreements (or not) on personalised advertising need to be unbundled from the overall Terms of Service.
The two complaints that led to the ruling were both made in 2018, on the first day that GDPR came into force.
More than 1,400 fines have been introduced since (GDPR) took effect, but this time the bloc’s regulators have shown they are willing to take on the very business model that makes surveillance capitalism, a term coined by American scholar Shoshana Zuboff, tick. “It is the beginning of the end of the data free-for-all,” says Johnny Ryan, a privacy activist and senior fellow at the Irish Council for Civil Liberties.
The regulators haven’t told Meta what they need to do to comply, but observers think that sooner or later (once the company has gone through the usual Big Tech shenanigans of appealing, “PR bullshit”, suggesting non-compliant solutions, etc) it will have to implement some version of Apple’s privacy change in 2021 that explicitly asks individuals if they want to be tracked. That’s quite a useful comparison, because we have some relevant numbers about that:
Meta told investors Apple’s move would decrease the company’s 2022 sales by around $10 billion. Research shows that when given the choice, a large chunk of Apple users (between 54 and 96 percent, according to different estimates) declined to be tracked. If Meta was forced to introduce a similar system, it would threaten one of the company’s main revenue streams.
The article suggests that this is part of a wider move away from largely unregulated online advertising, which is also affecting Google. Both Google and Apple introduced some technical changes to their advertising after GDPR was introduced. Max Schrens, the Austrian privacy activist whose NGO NOYB filed the two complaints against Facebook, as it was in 2018, says that “the combination of technical and legal shifts (are) moving in the same direction.”
My reading of this is slightly different: I’d say that the legal shifts are shepherding Big Tech businesses into changing their business models. One of the problems of dealing with GDPR is that fines for non-compliance can keep piling up.
Google, meanwhile, is trying to move away from cookies, although has delayed this, partly because of question marks about its approach, partly because some of the publishers it works with says that the abolition of cookies will harm their business models. They may be wrong about this, but once a junkie... (I wrote about this here a couple of years ago).
And of course, there’s nothing like a business model challenge to bring out the worst in the corporate affairs team of Big Tech companies:
Despite Google’s planned move away from cookies, the company has claimed that ditching personalized advertising altogether would jeopardize the authority of information online.
At least it hasn’t fallen back on that Big Tech stalwart, that any form of regulation will have “a chilling effect on innovation.”
What’s likely to happen eventually is an opt-in online culture:
“If everything becomes opt-in in the future, I think we have gained a lot because then we will actually have to understand what we’re opting into,” says Pernille Tranberg, cofounder of Danish think tank Data Ethics EU. Tranberg is not against personalized advertising, but she wants to choose which sites she gives her data to, depending on their reputation—she probably wouldn’t give her data to Facebook, she says, but she might give it to a newspaper or a bookstore.
From a business point of view, I’d say this was a good thing. It means that corporate reputation, behaviour, and brand are worth something, and therefore worth investing in.
One of the options that Google has been working on is ‘contextual advertising’, which is tracker-free and involves associating ads with the content that is being viewed. So a Volkswagen ad gets served when you’re reading an article about cars, for example. This seems more like the traditional print model, at least when it comes to specialist media. If I open a cycling magazine, I expect to see advertising related to cycling.
The Wired article suggests that this won’t seem so different from personalised advertising. But clearly it will. For one thing, ads won’t chase us across the web anymore. And Meta will find it a lot harder to accumulate information about its users.
(H/T The Overspill, I think.)
2: Surviving in wicked learning environments
In Monday’s Just Two Things, I posted the first part of a long review of ‘Range’, by David Epstein. Epstein argues in the book that in ‘wicked’ environments, where patterns may not be clear and feedback is delayed or incomplete, ‘breadth’ (or range) matters more than specialisation or depth. Here’s Part Two.
In some of the discussion in Range Epstein is fairly nuanced. He spends time, for example, on the dispute between Paul Erlich, the author of Population Bomb, and the economist Julian Simon, who thought that innovation would prevent Ehrlich’s bleak Malthusian world from happening. (So far it has).
This was, he suggest, a sterile discussion between two hedgehogs unable to see beyond their own disciplines:
Ehrlich was wrong about population (and the apocalypse), but right on aspects of environmental degradation. Simon was right about the influence of human ingenuity on the food and energy supply, but wrong in claiming that improvements in air and water quality also vindicated his predictions... As each man amassed more information for his own view, each became more dogmatic. (218)
He also dives into the briefly popular notion of ‘grit’, developed by Angela Duckworth. He acknowledges that he’s attracted to the idea because of his own career as a university athlete, where grit helped him overcome the limits of his natural talent.
But he concludes that the ‘grit’ questionnaire doesn’t help. It values persistence over everything, whereas in practice it’s more valuable to know when you have lost your ‘match’ with your environment, and can therefore decide to move on.
I lost patience with him a bit while reading a disguised eight page version of van Gogh’s life and work during his 20s, when he was an unsuccessful picture dealer and had a difficult period as a pastor.
Gaugin—who didn’t take up painting until he was 35–gets a call-out around here as well.
At the end of all of this—and it does go on—he pronounces with a flourish that
they aren’t exceptions by virtue of their late starts, and those late starts did not stack the odds against them. Their late starts were integral to their eventual success. (128)
But, but, but: Picasso, Monet, Matisse, Turner, Malevich, Kandinsky, even Cezanne (whose family required him to study law for two years) all contributed as significantly to innovation in modern art, while having much more conventional art trainings.
I had the same sense of cherry picking in a subsequent chapter about the persistence of Jill Viles, who suffered from a gene mutation, in alerting scientists to a way to understand better the lamin gene. Her persistence is remarkable, even gritty. Sure, but again, all post-hoc: for every Jill Viles there are 10,000 amateur medical researchers who are just plain wrong.
This isn’t to say that hedgehogs aren’t necessary. One of Tetock’s super-forecasters says she mines the domain experts for facts, but stays away from their opinions. Einstein was a hedgehog, which probably helped him to formulate the Theory of Relativity in the first place, but may have limited his later research:
God does not play dice with the universe, Einstein asserted, figuratively. Niels Bohr, his contemporary who illuminated the structure of atoms (using analogies to Saturn and the solar system), replied that Einstein should keep an open mind and not tell God how to run the universe. (229)
There is a couple of interesting conclusions here. One is about learning. In Tetlock’s super-forecaster study, he found that both foxes and hedgehogs were quick to update their beliefs after a successful prediction, reinforcing them. But:
When an outcome took them by surprise, foxes were much more likely to adjust their ideas. Hedgehogs barely budged. Some hedgehogs made authoritative predictions that turned out wildly wrong, and then updated their theories in the wrong direction. They became even more convinced of the original beliefs that led them astray. (231)
Put like this, of course, it sounds very like Leon Festinger’s work on cognitive dissonance.
What should you do if you find yourself in a wicked environment where you are, as it were, suddenly out of your depth? Epstein has a chapter called ‘Learning to Drop Your Familiar Tools’, which is both a factual description and a metaphor. It draws on the work of the organisational psychologist Karl Weick on smoke-jumpers—the specialists who fight forest fires. When situations got out of hand, often because a fire jumped a fire break, the people who survived were more likely to be the ones who, literally, drop their tools, their backpacks or chainsaws, to give themselves a better chance of outrunning the fire. But obviously there’s an analogy here as well.
A tweet in response to the Nature paper referenced above suggests eight ways to improve diversity in research thinking, some of which are also mentioned by Epstein. They are captured in a slightly enigmatic graphic.
There’s clearly a significant policy implication here. If we need more breadth and less depth—which is also the clear implication of the Nature article referenced above—we need universities and research institutes to value inter-disciplinary expertise as well as deep disciplinary knowledge. Reading this part of the book, I realised why futures work—definitely a realm of foxes or birds and not hedgehogs or frogs—has struggled to establish itself in academe.
One of my favourite sections of the book was about the Nintendo engineer Gunpei Yokoi, who transformed Nintendo’s business. Yokoi was by his own admission not a very good engineer, but he was a curious tinkerer. His operating principle for Nintendo’s innovation was “lateral thinking with withered technology”, meaning the technologies that were now so familiar that competitors were moving on from them. The Game Boy, which sold 118 million units, was the highlight of this approach:
From a technological standpoint, even in 1989, the Game Boy was laughable. It cut every corner... What its withered technology lacked, the Game Boy made up for in user experience. It was cheap. It could fit in a large pocket. It was indestructible. (196-197)
If there’s a paragraph that summarises the overall conclusion of Range, it comes about two-thirds of the way through the book:
Facing uncertain environments and wicked problems, breadth of experience is invaluable. Facing kind problems, narrow specialization can be remarkably efficient. The problem is that we often expect the hyperspecialist, because of their expertise in a narrow area, to be able to magically extend their skill to wicked problems. The results can be disastrous. (213)
Range is, in short, full of good things. It covers a lot of ground and a lot of research. All the same, it’s still about twice as long as it needs to be. My tip is to skip-read the extended narrative sections.
j2t#413
If you are enjoying Just Two Things, please do send it on to a friend or colleague.