With a bonus story: an AI system focused on the intelligence industry, identifying information campaigns, now spilling over into the commercial sector
30 October 2020 (Crete) – Last night I published a rough draft of the introductory chapter to my new monograph (due out in January entitled The struggle (futility?) of controlling or regulating technology in the modern world. I am pleased to say the response has been good. As of this morning I’ve received 109 comments that include some trenchant critiques for which I am appreciative.
My central premise is simple: the regulatory state must be examined through the lens of the reconstruction of the political economy: the ongoing shift from an industrial mode of development to an informational one. Regulatory institutions will continue to struggle in the era of informational capitalism they simply cannot understand. Regulatory processes are befuddled by the regulatory issues and problems created by information markets and networked information and communications technologies. That first chapter is a bit of a “TL;DR” but hopefully some of you will take the time to plow through it.
I have devoted a chapter solely on European regulation of Big Tech and thought I’d share some preliminary thoughts. There has been a lot of activity of late so I thought I’d focus on a few of the larger stories.
The EU’s absence in any element of the digital age’s data architecture is by choice, not necessity, as I detail in the monograph. The EU could have carved out a niche for itself. France and Germany still have some of the best mathematicians and engineers in the world. But politics intruded. The EU chose to prioritize data protection over technology (especially artificial intelligence) investment. For 15 years the EU has served as the world’s leading digital police force, making up for its lack of massive tech companies by wielding tools such as multibillion-dollar antitrust fines, tax claw-backs and laws like the General Data Protection Regulation (GDPR).
And the EU Commission remains perplexed why Big Tech behavior “has not changed, even after we have levied all these fines” – the thought recently expressed by Margrethe Vestager in an unguarded moment. She has dual roles: the EU’s Competition Commissioner, and also as Commission Executive Vice-President for dealing with Europe and the Digital Age.
Yes, European regulators have attempted to articulate a more demanding conception of permissible market behavior, at least contending more directly with the various kinds of external costs that platform power can create. But whether or not they’ll be effective remains to be seen. The EU has levied a lot of fines … chump change for these companies … but almost zero change in behavior. Frankly, despite all the “digital age” talk, I think European regulators are still in thrall to the analogue mindset. The five-year digital policy blueprint the EU Commission unveiled this year was written as if the technology age will not develop any further.
You see it in the GDPR, both the wording of the Regulation and its implementation. As I have written before, the GDPR will ultimately fail because it is based on three fallacies: (1) the delusion that data protection law can give individuals control over their data; (2) the misconception that the reform simplifies compliance, while in fact it makes compliance even more complex; and (3) the assumption that data protection law should be comprehensive, stretching data protection to the point of breaking which EU regulators are now recognising. Data protection specialists and information governance mavens do not want to talk about this because they cannot make any money from it. Their task is to sell data protection service and product “compliance solutions”.
And EU regulators are lazy sods. In the first chapter of my monograph you can read about the Volkswagen emission scandal which illustrates the pervasive institutional influence of economic power — and shows that influence operating on levels that are both political and ideological. In the weeks after the news broke, press coverage documented Volkswagen’s systematic efforts to stave off more intrusive regulation in the European Union and probed the close ties with the private European emissions testing laboratories that acted as EU regulatory surrogates – and the EU regulators were happy to acquiesce.
There have been 200+ articles on EU competition law and Big Tech over the last few months. Here are two recent stories that illustrate how things are going:
Amazon and Apple are being probed by German regulators over some interesting online sales curbs
“Brandgating agreements can help to protect against product piracy. But such measures must be proportionate to be in line with antitrust rules and may not result in eliminating competition.”
Authorities are wrestling with how to act against companies that critics say run a rigged game when they set the rules for platforms that also host their rivals.
As I noted in my monograph piece, “gaming” the regulatory system is the thing to do for all these companies as I quoted from a long-time friend at Oligvy (the company that provided my first job in the advertising business). From his longer quote: “In markets for information-related goods and services, consumer awareness is easy to manipulate more directly, and the goods and services frequently are amenable to versioning in ways that embed material nonprice terms within price discrimination frameworks. Yes, we game the system all the time. It’s easy. Especially goods and services that have latent, complex or highly technical characteristics that consumers cannot fully understand or value accurately. Room to play.”
Yep. Power imbalances and other structural imbalances will undercut or frustrate efforts to obtain more comprehensive and accurate information.
Of course Amazon said in a statement that it never removes sales permissions without a sound reason and invests heavily to protect customers from the illegal distribution of goods. It said it’s cooperating with the regulator. But Apple is a prominent example of how Amazon does brandgating, which can take various forms. Since the start of 2019, only authorized Apple vendors can sell via Amazon’s marketplace. Amazon itself became such an authorized seller.
The Facebook antitrust data-hunt gets ground rules
In a decision that isn’t a decisive victory for either side, the president of the EU’s second-highest court (the European General Court) said on Thursday that the European Commission can’t force the social network giant to hand over potentially sensitive records without a detailed review.
Instead, the EU must work with Facebook in identifying such information and store it in a “virtual data room” (what the hell IS that?). The ruling follows a dispute over arrangements on how to view the data.
“The members of the team responsible for the investigation shall examine and select the documents in question,” while giving Facebook lawyers “the opportunity to comment on them before the documents considered relevant are placed on the file”. You’ll find the order here. And if you Google “T-451/20 and T-452/20 Facebook Ireland v Commission” you’ll find loads of stuff.
The EU last year started examining Facebook’s sales platform and how it uses data from apps, as part of a broader crackdown on Silicon Valley. The EU is also looking at how Amazon collects data from retailers through its platform and investigations into Apple app store. Regulators can require companies to give documents mentioning certain keywords under threat of fines. Facebook sued the commission in July, citing “the exceptionally broad nature” of the EU’s orders. It also filed two challenges seeking a court suspension of the EU data demands.
Facebook said in a statement that it particularly welcomed the court’s assessment “that highly personal and irrelevant information enjoy strong legal protections which need to be respected in the commission’s ongoing investigation. In the meantime, we continue to cooperate with the commission and have already provided it with over a million documents.”
And, of course, the EU Commission said it “will duly implement the court’s order and continue to defend its case in court. The commission’s investigation into Facebook’s potential anticompetitive conduct is ongoing.”
The EU’s demands impose on Facebook “a positive obligation to search for all of its electronic files on the basis of broad search terms and to communicate to the commission the documents responding to those search terms, even if those documents contain sensitive personal data,” the court decided. Facebook “correctly claims that that obligation requires it to process sensitive personal data.” The decision on so-called interim measures can be appealed. Facebook’s main appeals over the legality of the EU’s demands remain pending.
And my bonus feature because I did not know where to put it …
Recent leaps in machine understanding of language, enabled by feeding large machine-learning models huge amounts of text training data, have had a big impact on the intelligence community as well as business. Last year I wrote about a company called Primer which I met with at the International Cyber Security Forum in Lille, France. Primer is focused on the intelligence industry, but it started to go “commercial” having won a contract two years ago to supply its technology to Walmart, for identifying buying trends and supply chain issues. It is also being tested for use by an e-discovery/information governance vendor (an NDA prevents me from disclosing the name) so things should get interesting next year.
Primer surfaced again this week during a Zoom chat I was invited to monitor (“listen, don’t talk, take notes”) which concerns the much-in-the-news fighting between Azerbaijan and Armenia fighting over Nagorno-Karabakh, a disputed territory in the Caucasus mountains. There has been a massive information warfare campaign over the region had been underway for several months. The campaign was identified using artificial intelligence technology being developed for US Special Operations Command which oversees US special forces operations. That AI system was powered by Primer
The AI can identify key themes in the information campaign by analyzing thousands of public news sources. In practice, Primer’s system can analyze classified information too. The analysis in the Zoom call showed how Russian news outlets began pushing a narrative in July designed to bolster its ally Armenia and undermine its enemy Azerbaijan. The U.S. Department of Defense faces a torrent of information as well misinformation. To keep pace, you’re not going to do it with a bunch of people reading. This has to be machine-enabled.
The e-discovery vendor I noted above is testing it for unstructured data bases, and various non-static databases and targets.
The defense and intelligence industries need to analyze an avalanche of unclassified information such as social media along with classified reports, creating opportunities for companies like Primer, Splunk, Redhorse, and Strategic Analysis. This increasingly involves drawing insights from different forms of data such as voice recordings and images as well as text.
Primer’s analysis of the Russian information campaign offers a simple example of how AI can both help organize information and identify misinformation. The effort to portray Azerbaijan and Turkey as aggressors in Nagorno-Karabakh may have provided an early warning sign of escalating tensions, or an indicator of Russia trying to stir up trouble. The report boiled down 985 incidents from over 3,000 documents to 13 key events. A human analyst would have needed several hours to perform the analysis, but the Primer system did it in roughly 10 minutes. The technology works across several languages, including Russian and Chinese as well as English.
And I know there’s been plenty of hype around AI for defense, but most of the participants in this Zoom chat were of common mind: “We’re past the snake oil phase now. They seem to have pretty sophisticated natural language capabilities. One can only imagine what this will mean for the commercial sector when the really sophisticated stuff we use is released to the common”.
I think it also provides a good example of how analysts (military and commercial) will work together with AI systems in the future. This tech is not mind-blowing in the sense of an AI that can understand everything, but what Primer is doing is the toughest technical challenge in machine understanding of language and it’s pretty impressive.