Google's Spamgate

Episode 405 of the Cyberlaw Podcast

Retraction: An earlier episode of the Cyberlaw Podcast may have left the impression that I think Google hates mothers. I regret the error. It appears that, in reality, Google only hates Republican mothers who are running for office. But to all appearances, Google really, really hates them. A remarkable, and apparently damning study disclosed that during the most recent federal election campaign, Google's Gmail sent roughly two-thirds of GOP campaign emails to users' spam inboxes while downgrading less than ten percent of the Dems' messages. Jane Bambauer lays out the details, which seem to refute most of the excuses Google might offer for the discriminatory treatment. Notably, neither Outlook nor Yahoo! mail showed a similar pattern. Tatyana thinks we should blame Google's algorithm, not its personnel, but we're all eager to hear Google's explanation, whether it's offered in the press, before the Federal Election Commission (FEC), in court, or in front of Congressional investigators after the next election.

Jordan Schneider helps us return to China's cyber policies after a long hiatus. Things have not gotten better for the Chinese government, Jordan reports. Stringent lockdowns in Shanghai are tanking the economy and producing a surprising amount of online dissent, but with Hong Kong's coronavirus death toll in mind, letting omicron spread unchecked is a scary prospect, especially for a leader who has staked his reputation on dealing with the virus better than the rest of the world. Among the results is hesitation in pursuing what had been an aggressive techlash regulatory campaign.

Tatyana Bolton pulls us back to the Russian-Ukrainian war. She notes that Russia Is not used to being hacked at anything like the current scale, even if most of the online attacks turn out to be pinpricks. She also flags Microsoft's report on Russia's extensive use of cyberattacks in Ukraine. All that said, cyber operations remain a minor factor in the war.

Michael Ellis and I dig into the ODNI's intelligence transparency report, which inspired several different takes over the weekend. The biggest story was that the FBI had conducted "up to" 3.4 million searches for U.S. person data in the pool of data collected under section 702 of the Foreign Intelligence Surveillance Act (FISA). Sharing a brief kumbaya moment with Sen. Ron Wyden, Michael finds the number either "alarming or meaningless," probably the latter. Meanwhile, FISA Classic wiretaps dropped again in the face of the coronavirus. And the FBI conducted four searches without going to the FISA court when it should have, probably by mistake.

We can't stay away from the pileup that is Elon Musk's Twitter bid. Jordan offers views on how much leverage China will have over Twitter by virtue of Tesla's dependence on the Chinese market. Tatyana and I debate whether Musk should have criticized Twitter's content moderators for their call on the Biden laptop story. Jane Bambauer questions whether Musk will do half the things that he seems to be hinting. I agree, if only because European law will force Twitter to treat European sensibilities as the arbiter of what can be said in the public square.

Jane outlines recent European developments showing, in my view, that European policymakers aren't exactly running low on crazy. A new EU court decision opens the door to data protection class actions, undermining the jurisdictional limits that have made life easier for big U.S. companies. I predict that such lawsuits will also mean trouble for big Chinese platforms.

And that's not half of it. Europe's Digital Services Act, now nearly locked down, is a mother lode of crazy. Jane spells out a few of the wilder provisions – only some of which have made it into legal commentary.

Orin Kerr, normally a restrained and professorial commentator on cyber law, is up in arms over a recent 9th Circuit decision holding that a preservation order is not a seizure requiring a warrant. Michael, Jane, and I explore Orin's agita, but we have trouble sharing it.

In quick hits:

• Jane makes short work of a report expressing shock that Amazon uses data from Alexa smart speakers pretty much exactly the way you'd expect it to.
• Michael and I unpack the latest move in the prosecution of Uber's former Chief Security Officer, Joe Sullivan.
• Jane lays out what's different in Colorado's new privacy law. Spoiler:  Just enough to make a federal privacy law with preemption look good to business.
• Michael and I wish the Biden administration well in its effort to get much-needed new authorities to address the risks of drone attacks here at home.

Download the 405th Episode (mp3)

You can subscribe to The Cyberlaw Podcast using iTunes, Google Play, Spotify, Pocket Casts, or our RSS feed. As always, The Cyberlaw Podcast is open to feedback. Be sure to engage with @stewartbaker on Twitter. Send your questions, comments, and suggestions for topics or interviewees to [e-mail]. Remember: If your suggested guest appears on the show, we will send you a highly coveted Cyberlaw Podcast mug!

The views expressed in this podcast are those of the speakers and do not reflect the opinions of their institutions, clients, friends, families, or pets.

Comments

[Brett Bellmore]

"I agree, if only because European law will force Twitter to treat European sensibilities as the arbiter of what can be said in the public square."

And that's something we need to put a stop to, ASAP. This is a confrontation that needs to take place ASAP, and be fought to the bitter end.

[apedad]

???

Who do you want to fight and over what?

If Twitter wants to adopt the European standards then. . . so?

[Brett Bellmore]

The law in question isn't about what Twitter wants to do. It's about what Europe proposes to force Twitter to do, using the threat of massive fines. Based not on revenues in Europe, by the way, but world-wide revenues.

There are actually positive aspects of the European law, but then there's the “legal but harmful” clause, which category is still up to be defined, but sure looks like it's going to be highly political.

[Sarcastr0]

...You want America to invalidate European laws?

Sovereignty can suck it, I guess.

[Brett Bellmore]

No, I want Europe to go pound sand.

OK, so Europeans don't have constitutionally protected free speech rights, in the UK they don't even HAVE a constitution. I don't want that exported here on the basis of Europe threatening massive fines if a US company doesn't assist them in oppressing their own people.

Put the IT infrastructure outside of Europe, and then let them go pound sand. If they want to build a European equivalent of the Great Firewall of China, let them do it themselves. No reason a US company should give a damn about European laws if somebody from Europe logs into a US service.

Of course, there's a veiled threat they'll go after Telsa if Musk doesn't knuckle under.

[DaveM]

The study of bias in email spam filters looks really solid. I was genuinely surprised by how significantly Gmail in particular responds to user interaction with email marked as spam. My key takeaway: don't blindly trust spam filters. Review your spam folder from time to time and unmark the false positives, especially if you use Gmail.

[Michael P]

Review your spam folder from time to time and unmark the false positives, especially if you use Gmail.

Very much this. My experience with a Google email account (not Gmail proper, but through Google Apps or Workspace or whatever they call it this year) suggests that the bias is not necessarily explicitly set up, for two reasons:

A) It reliably puts mass mailings from Democratic campaigns into my spam box. (Someone must have thought it would be funny to tell Democrats that I was a good prospect for them.)

B) It very frequently puts mail from the "golang-nuts" mailing list, run by Google to discuss Google's Go programming language, into my spam box. I finally tweaked my filter for that to tell it to never mark that mailing list as spam.

So I suspect the bias is probably real, but probably also some kind of emergent behavior based on how users classify spam.

[Brett Bellmore]

I would not be particularly shocked if Gmail's spam identification algorithms were biased, Google has been caught politically biasing their search results.

But neither would I be shocked to find that a good deal of it was just neutral application to different political email practices. Frankly, I think a lot of the political email I get from Trump, and sites that want me to think they're Trump, ARE spam. And I voted for the guy!

[reasonable]

I'm not sure that a difference in outcomes is necessarily "bias" or at least intentional bias. While the exact algorithms are opaque, one large factor is when users mark an email as spam. Each time that happens, the algorithms get trained to believe that the keywords in the messages and the senders' email domains are more likely to be spam. If more Google users mark Republican emails as spam, the algorithm starts interpreting any email containing said keywords and/or sending email domains as more likely to be spam. If more Google users reply, forward or interact (click on a link) with Democratic emails, the algorithms get trained to interpret those email has being not spam. Further, many platforms are trying to combat "hate speech" and debunked conspiracy theories. If Republican emails contain more hate speech and conspiracy theory related terms, the more likely they are to be marked as spam.

[Brett Bellmore]

"Further, many platforms are trying to combat "hate speech" and debunked conspiracy theories. If Republican emails contain more hate speech and conspiracy theory related terms, the more likely they are to be marked as spam."

But there you're back to intentional bias, because the definition of 'hate speech' is politically biased. Like, noticing that the Assistant Secretary for Health is a guy? "Hate speech".

[RightWired]

"Researchers find Amazon uses Alexa voice data to target you with ads"

My wife and I have had the same amazing Dyson vacuum for 10+ years. It's never caused any issues other than a faulty power switch, which I fixed itself - we have never looked for a new one.

Last week, the power cord became exposed. One breakfast morning in front of our Alexa I said "We may need a new vacuum"...wife said "what about that Shark? Mark and Susan got one".

The next day, I was seeing Shark Vacuum ads. Neither of searched for one. Just our casual conversation. It's clear: Bezos is spying

[Brett Bellmore]

My 13 year old son, who long thought I was just paranoid about Alexa listening it, just recently confessed to me that he believed me, because he'd been talking with a friend in front of the thing about a video game he wanted, and, presto! He starts getting ads for it.