The Unwisdom of Crowds (and Hypernetworked Tech VCs)

What last week's banking failures can teach us about the dangers of our high-speed information economy. Plus how rightwing Catholics are weaponizing mobile app data.

“ON MONDAY 100,000 AMERICANS WILL BE LINED UP AT THEIR REGIONAL BANK DEMANDING THEIR MONEY — MOST WILL NOT GET IT….THE FARMER IS THE SMALL BUSINESS OWNER WHO WONT BE ABLE TO MAKE PAYROLL NEXT”--Angel investor, “This Week in Startups” podcaster, and Elon Musk crony, Jason Calacanis, on Twitter, Saturday, March 11, 2023

“This was the first Twitter fueled bank run.”--House Financial Services Committee Chair Patrick McHenry (R-NC), March 12, 2023

“This was more a case of a ‘bank-run by idiots’ rather than a ‘bank run by idiots’.”--Financial journalist Matthew Klein, March 13, 2023

There are two big lessons to be learned from the sudden collapse this weekend of Silicon Valley Bank and Signature Bank, two mid-sized institutions that catered heavily to tech start-ups. The first is an old one: private capital sits in the driver’s seat of the economy and always gets preferential treatment (like loosened regulations followed by inevitably necessary bailouts, or in this case, backstops). Until we better organize countervailing power—building things like labor unions, debtors unions, and public pension funds—and prove to politicians that they have more to fear from us than them, this will keep happening. Computerized global capitalism (plus mobile banking apps that save you a trip to the bank to demand your funds) just amplifies the risk and scale of these disasters.

As financial historian Adam Tooze put it earlier today, “On the face of it SVB is not a bank big enough to do systemic damage. That was the prima facie reason for exempting it from the oversight that extends to really big banks. Finance per se will not explain the emergency intervention. But as should have been obvious all along SVB matters very much indeed, because its depositors are very powerful, very rich and very influential people who own a narrative that makes them indispensable to one vision of America’s future. And that force was brought to bear on the Biden administration over the weekend in an extraordinarily overt exercise of ‘venture dominance’.”

He goes on to describe how the National Venture Capital Association gathered a thousand participants in a Zoom call Friday night with a top Treasury official, where speakers claimed that if the mega-depositors who had foolishly concentrated their holdings at SVB weren’t made whole, the financial system could collapse. Tooze adds, “This is what it looks like when the bourgeoisie in the true sense swings into action. It is what it looks like when an executive committee or committees constitute themselves and demand action from the state.” (Here’s a gift link to Teddy Schleifer’s revealing report for Puck News on how one of Silicon Valley’s best-connected VCs, Ron Conway, rallied people like former president Barack Obama, former House Speaker Nancy Pelosi, and current Vice President Kamala Harris, to the rescue brigade.)

Seeing MAGA Republicans blaming SVB’s collapse on excessive “woke-ism” at the bank because it had a relatively diverse board is just icing on the cake from the GOP distraction machine. All of these politicians supported rolling back the Dodd-Frank rules overseeing mid-size banks like SVB back in 2018 (along with the handful of Democrats who also ought to be remembered for voting for deregulation, like Rep. Sean Patrick Maloney and Rep. Josh Gottheimer).

The second lesson isn’t new either: our systems for building and deploying group intelligence are quite vulnerable to herd behavior when fear strikes. Our collective amygdala is too easily hijacked. But unlike bank runs, which have been happening for centuries and which we’ve learned to tame, social media panics are relatively new. While it’s truly a great thing that the world is wired together like never before, we need to figure out where to put some circuit breakers. I’m not suggesting that the run on Silicon Valley Bank wouldn’t have happened if Elon Musk hadn’t fired the entire Twitter Trust and Safety team months ago; tech VCs and start-up CEOs did a good enough job of panicking each other in private chat groups even before idiots like Jason Calacanis and his buddy David Sacks fanned the flames in public. But this weekend’s mini-panic shows again that online systems designed to optimize engagement (and profit) can spiral quickly and dangerously.

Wikipedia, the one giant online institution that has avoided this problem, manages to work because a small army of community volunteers vigilantly guards the site from disinformers and conflict entrepreneurs (see image above for an example of how it does that). On the other hand, other big pieces of our global brain, like Facebook and YouTube, do very little to slow the velocity or guard the veracity of what flows through their services. Sure, when it became clear that WhatsApp chat groups in south Asia were becoming vectors for spreading vigilante violence against some Muslim communities, Meta executives did choose to limit how often a WhatsApp post could be forwarded. That’s a kind of circuit-slower, if not a circuit-breaker. But voluntary action by big tech platforms is a very thin reed to rely on.

This should be of even greater concern as we’re now in another hype cycle driven by tech VCs unfettered capitalism, this one not being about “big data” or blockchain or Web3, but about something quite powerful: generative artificial intelligence. Tools like ChatGPT and DALL-E clearly have uses, but none of the companies developing them have any incentive to slow down and consider the dangers they may unleash, since they all believe if they do one of their competitors will beat them to the market.

And we shouldn’t wait for another financial panic fueled by digital media to build some countermeasures. Lots of places in America have been having mini-panics for a while now, over everything from rumors of antifa buses on their way to fears that “groomers” are stealing their children. I realize this is a bit like shutting the barn door after the lady who thinks Jewish space lasers set off California forest fires got embraced by the Republican House Speaker. But in the same way that towns and communities practice for “civil defense,” we need to invest in collective information defense too. Or else a carnival barker like Jason may yet see his worst fears fulfilled.

Bonus link: Paris Marx (who writes the excellent

Disconnect

substack) has a fine rant on this topic titled, “The Silicon Valley Bank Collapse Should be a Radicalizing Moment.”

Privacy, Shmivacy

Last Thursday’s story (gift link) in the Washington Post by Michelle Boorstein and Heather Kelly on a conservative group of Colorado Catholics that used mobile app tracking data to covertly investigate and expose priests using gay dating apps hasn’t gotten enough attention for what it portends. Though some tech journalists and privacy groups have been warning for a while now that aggregated and supposedly anonymized digital data can be reverse-engineered to expose individual users, this is the first time that an advocacy organization—as opposed to law enforcement—has done this.

According to the Post’s story, the group, Catholic Laity and Clergy for Renewal, obtained data spanning 2018-2021 from multiple dating sites mostly catering to gay men (Grindr, Scruff, Growlr and Jack’d) along with OkCupid, which caters to all. “The group cross-referenced location data from the apps and other details with locations of church residences, workplaces and seminaries to find clergy who were allegedly active on the apps,” Boorstin and Kelly write. Though they got no names from data brokers, they were able to zero in on individuals by focusing on devices that spent multiple nights at a rectory, or if a hookup app was used multiple days in a row from a church building. Then it shared its findings and suspicions with various bishops around the country. It appears that Renewal’s data-mining led to the July 2021 outing of a prominent priest, Monsignor Jeffrey Burrill, from a top position at the US Conference of Catholic Bishops.

Jayd Henricks, the president of Catholic Laity and Clergy for Renewal (and former executive director of government relations of the US Conference of Catholic Bishops), is defending this new kind of digital witch-hunting, posting a response on First Things, a leading Catholic magazine, a day after the Post story. He writes that his group formed after a 2018 scandal involving a Cardinal who had been sexually abusing young men for decades. He writes:

“The purpose was simple: to love the Church and to help the Church to be holy, with every tool she could be given. And of all the means the Church has employed to help herself, addressing the role of technology was a glaring omission, particularly given that Pope Francis has called for the use of technology for the common good.

Based on that insight, a group was formed, Catholic Laity and Clergy for Renewal (CLCR), to explore ways technology might serve the bishops in addressing their greatest challenges. I’ve been proud to be a part of that group.  

After all, data is used by all major corporations, so why not the Church?”

He says they’ve done data-driven studies on why Catholics leave the Church and how they engage across social media, for example. And he doesn’t deny snooping into priests’ private lives. “When we learned legal ways to understand risks to the health of the Church beset by technology—including the use of hookup apps by clerics—we studied that,” he writes. “We learned some things. And we shared what we learned directly with bishops—without setting any expectations, we made information available to the leaders of the Church.”

Henricks goes on to argue that his group only focused on “behavior that harms everyone involved” and not on whether or not a priest or seminarian was straight or gay. And he insists that if they had discovered “any illegality, like abuse of minors” it would have been reported to law enforcement. And he insists that all he is doing is trying to help the Church he loves produce “faithful and joyful pastors.”

That’s one way to interpret this. Or you can take Boorstin and Kelly’s interpretation: “The project’s existence reflects a newly empowered American Catholic right wing that sees enforcing its interpretation of church teaching on sexuality and gender as an existential issue for the church and that no longer trusts bishops to do so. It is a flip of traditional church power dynamics, with the Colorado laypeople in a position to pressure bishops. At the most intimate level, it shows a new generation of surveillance technology moving into different realms, now including the religious.”

The scandal, also, is what’s legal. Though all of the hookup sites mentioned in the story say they have since tightened up how they share data with third-parties, buying and selling of location data is still prevalent in the digital ad industry. There’s no law stopping anyone from using location data to target or expose individuals. This is the world we live in.

Odds and Ends

—Tim Chambers of the Dewey Square Group is out with a useful report on “The Twitter Migration,” which he says is continuing in spurts as the chaos at Twitter spikes, with most of the action flowing to the Mastodon fediverse, which is seeing about 2.4 million monthly active users, up from 300,000 a month before Twitter was acquired by Musk.

—Meta, the oh-so-innovative company used to be known as Facebook, is reported to be developing a Twitter competitor codenamed P92 that would be built on the same tech protocol as Mastodon, Deepsekhar Choudhury and Vikas Sn report for MoneyControl.

—GPT-4 is here and apparently it’s good at evaluating and responding to images. And it does very well on SATs. And Khan Academy is already using it  as a virtual tutor for students. And judging from a live OpenAI demo I just watched, it can read the tax code and help you do your taxes.

—In last week’s Connector, I speculated about the use of speech cloning tools to dupe people into believing their loved ones were in need of urgent help. It turns out that’s already happening, as Pranshu Verma reports for the Washington Post. In one case, parents received a phone call from someone claiming to be a lawyer representing their son, who had supposedly killed a US diplomat in a car accident. Then the “son” got on the phone and said he loved his parents and needed the money. They wired more than $15,000 to the “lawyer” before discovering they’d been scammed.

—Gideon Lichfield, the editorial director of Wired magazine (and in my opinion one of the sharpest editors on the tech and society beat) has shared an initial set of guidelines for how the publication will (and won’t) use generative AI tools. This seems like an excellent transparency practice for organizations to follow.

—“If a political candidate can send more personalized messages using generative AI, surely they will.” That’s GWU political science professor Dave Karpf opining last Friday during a webinar on ChatGPT and politics hosted by the school’s Ethics in Political Communication program. He also said that we should expect to be flooded with deep fakes in the coming months, now that they’ve become so cheap and easy to make, and that we seem to be repeating the mistakes of the 1990s, when government chose to avoid regulating the commercial Internet “because we wouldn’t want to tamp down on these wonderful companies.”

End Times

Happy Pi Day!

I don’t ask all that often, but if you are getting value from these posts, please consider becoming a paying subscriber. I haven’t put up a paywall because I want this content to be as accessible as possible to all, but if you can afford a sub that helps keep the fires burning.