Surveilled—Issue 63: US elections, Bitcoin and AI bias
Surveilled
Issue 63
A weekly irregular summary of what I've found interesting at the intersection of economics, finance and technology.
Apologies for the longer than expected hiatus, I struggled to live up to the “weekly” part of the description due to Malaysia’s budget period, which is prime time for the various economic think thanks and media. If you’re at all interested in that, there is a selection of links below. The other big event in the period since the last issue was the US presidential election of course, which is covered in the first big story below.
Published Elsewhere
As mentioned, the early weeks of November for me were marked by the announcement of Malaysia’s federal budget for 2021, followed by a few weeks of analysis and debate. I appeared on a number of webinars, media and podcasts on the topic, selected links below:
- “Pre-Budget 2021: Usher in a new economic paradigm”, The Edge Weekly
- “Post Budget 2021: A New Economic Paradigm?”, REFSA (with former minister for International Trade and Industry Dr. Ong Kian Ming)
- REFSA Reflections podcast, “Budget 2021 with Ong Kian Ming”
- “Budget 2021: Overinvest in health”, Consider This, Astro Awani
- “Economists in the House: Analysing Malaysia’s 2021 Budget”, Gabungan Bertindak Malaysia (GBM)
- “Budget 2021 ambitious, but economic analysts say Putrajaya could be bolder with spending amid Covid-19”, Malay Mail
Big Stories
US Presidential Election: the Aftermath—From a global perspective the big story was of course the US presidential election, which among others constituted an acid test for the biggest social media platforms. On the whole the platforms did well, moving quickly to address false statements around election integrity etc., and likely contributing to settling the uncertainty around the results rather quickly.
There were notable differences in approach however, some platforms fared better than others, and more interestingly, the actions they took divulged more about their inner workings than what we already knew.
Facebook, for example, had to take emergency action to recalibrate its news feed to show more news from mainstream sources like CNN, the New York Times and NPR, instead of more partisan sources. This capability has apparently existed for a long time, but Facebook’s management opted against using it because the more partisan posts generate more engagement. Similar tensions were manifestly at work at Twitter and YouTube as well, although the former arguably managed the situation better than the latter.
Of course, the relatively calm aftermath of the election is not only thanks to the platforms themselves, but also to the government, in particular the Cybersecurity and Infrastructure Agency (CISA), whose director was fired by Donald Trump, supposedly for being too effective. While the firing in itself is not particularly interesting, the accompanying article provides an interesting glimpse at the agency and its environment.
Altogether though, the election hasn’t brought us any closer to a resolution for the big questions the social media platforms raise. If anything, their actions have underlined to what extent they control the narrative, and that their interests as commercial organisations tend to go against the interest of society as a whole.
We already knew this of course, and have been grappling with this issue for a while. Sadly the various attempts to contain the gravest excesses on the platforms in some countries have led to entirely unintended and unwanted consequences in other countries, as this example of Germany’s law against ‘fake news’ illustrates: the German law has been used as a template to stifle dissent in countries from Venezuela to the Philippines, over Russia and Singapore. Unfortunately, these externalities are intrinsic to the nature of the platforms as global commercial organisations, and it will take a long time to land on a model that works. Let’s hope that the damage done in the meantime will not be irreversible.
The Bitcoin Business Cycle—Bitcoin has been on a tear lately, closing higher than its previous high at the end of 2017. I don’t think Bitcoin can be considered as an asset, given that it has no intrinsic value or cash flows, but there is an argument to be made that there is a rationale for buying it as the ultimate doomsday hedge. On the other hand the fundamental issues with Bitcoin remain, and so far it seems to follow a fairly predictably boom/bust cycle. Buyer beware…
Six Links
One of the worrisome longer-term consequences of the pandemic is the vast increase in surveillance infrastructure, in particular in the workplace, where sales of contact tracing devices are booming. Read (OneZero)
The reason it is worrisome is because once the pandemic is under control, the available infrastructure will not be rolled back, and it will be too tempting to use it for other uses. Case in point, Amazon has already been spying on its employees working to organise labour unions, even hiring the Pinkerton Detective Agency, notorious for breaking steel worker strikes in the US in the 19th century. Read (Vice)
Max Schrems, the privacy activist who scored significant court wins against Facebook, is now targeting Apple, claiming they track users without consent through the unique device ID. Apple had already announced plans to make this more difficult, so it will be interesting to see how this goes. Read (FT $)
We worry a lot about bias in AI, but this story is a good reminder where that bias comes from in the first place: Dataminr is a company that basically searches Twitter for terms that might indicate gang activity, and turns over the results to the police. Read (The Intercept)
The evolution of AI research is not encouraging either. Instead of a widening ecosystem, where research from more diverse institutions and locations could help diminish built-in bias, AI research has been concentrating in the very largest firms and at the most established universities. Read (Arxiv)
And as if we didn’t have enough problems, researchers have designed a method to derive the contents of messages by tracking shoulder movements in Zoom calls. The results look a lot more accurate than what you’d expect them to be, so perhaps turn off video when writing a sensitive message. Read (Techxplore)
That's it for this week's edition. As always, thanks for reading and please forward this to anyone who you think might be interested, it would be much appreciated.