Good Friday Morning, except to Anthony Fauci, who had his emails released via FOIA requests this week. The Washington Post (~800 pages of emails) and Buzzfeed (~3200 pages of emails) combined to released around 4,000 total pages of emails from Fauci’s thoughts during the pandemic. Faucis’ book got scrubbed from booksellers’ websites, which made Rand Paul quip, “Oh, I don’t know. I think they should publish it. I love science fiction.”
The podcast will be back this week. I didn’t record over the Memorial Day weekend. It’s grown a lot lately, with three of the top five more listened-to episodes coming in the past five episodes. I thank you all for your support and look forward to building that out. This week, I’m going to go through why Facebook and social media can’t monitor speech in the way they claim and what that means for Congressional action in the future. Links to follow.
Where you can find me this week
Please subscribe, rate, and review my podcast on iTunes, Spotify, or Google Play — the reviews help listeners, and readers like you find me in the algorithms. Make sure to sign up for the Conservative Institute’s daily newsletter and become a subscriber at The Dispatch, where I’m a contributor.
Poppies, flags, and what we do to remember on Memorial Day – The Conservative Institute.
Yes, it’s important to know China’s responsibility on COVID-19 – The Conservative Institute.
BigTech vs. Politics.
At some point this week, Donald Trump’s pages on Instagram and Facebook reappeared. It does not appear he’s been reinstated to Facebook. One of Facebook’s communications people said, “No. Nothing about the status of President Trump’s presence on our platform has changed. He remains indefinitely suspended.” But you can comment and interact with old posts again. This move comes along with another set of announcements from Facebook on the treatment of politicians. The Verge had the exclusive report:
Facebook plans to end its controversial policy that mostly shields politicians from the content moderation rules that apply to other users, a sharp reversal that could have global ramifications for how elected officials use the social network.
The change, which Facebook is set to announce as soon as Friday, comes after the Oversight Board — an independent group funded by Facebook to review its thorniest content rulings — affirmed its decision to suspend former President Donald Trump but critiqued the special treatment it gives politicians, stating that the “same rules should apply to all users.” The board gave Facebook till June 5th to respond to its policy recommendations.
Facebook also plans to shed light on the secretive system of strikes it gives accounts for breaking its content rules, according to two people familiar with the changes. That will include letting users know when they’ve received a strike for violating its rules that could lead to suspension. BuzzFeed News and other outlets have previously reported on instances when Facebook employees intervened to keep political pages from being subject to harsh penalties under the strikes policy.
Facebook is also set to begin disclosing when it uses a special newsworthiness exemption to keep up content from politicians and others that would otherwise violate its rules.
I don’t care what you think about the wisdom of such a move. The reality is this: it’s going to end badly. There’s no way for Facebook or any tech company to monitor and police speech by anyone. And the idea that cracking down on politicians will help political discourse in this country is, at best, wildly naive.
I already see the far-left and progressives in my Facebook feeds chafing under the new rules. Facebook’s algorithms cannot discern all the subtleties of American English use, with its heavy use of hyperbole, sarcasm, snark, dark humor, and more. There are layers and sub-layers to American English that make it, in my mind, utterly impossible to police by any machine-learning algorithm.
Algorithms and why they can’t monitor speech.
My job as an attorney lets me dabble in this field. I deal with e-Discovery across large class action, mass tort, and other prominent cases. I regularly deal with situations where there are millions of documents involved. We have to figure out a way to get through those documents in a cost-effective manner. Artificial Intelligence and machine learning are the new tools that have grown over the last decade. While they’ve gotten a lot better, they’re still imperfect, and they rely heavily on human input to train them.
These tools aren’t something that you turn on, and they figure things out. We have a Marvel, Star Trek, or Star Wars sense of AI, where you turn it on, and it’s this thing that learns and develops like humans. But it doesn’t work like that. AI/machine learning only does what we, the humans, train it to do. It’s entirely possible — likely, in fact — for an AI system to reflect all the biases of humans.
For instance, a few years ago, Google had to shut down and apologize for an algorithm created to make labels for pictures in its database. Over time, the AI started labeling black couples as monkeys, and researchers couldn’t figure out why that was happening. During the pandemic, a second racist incident appeared with another algorithm that labeled darker skin tone hands that held thermometers as holding a gun. Police departments have had issues with predictive policing algorithms showing overly racist tendencies to over-police black neighborhoods more than others.
We create algorithms. We populate them with data. We make the parameters. An algorithm or AI program only does what we tell it with the data we give it. And I know, based on what happens in law and the issues we’ve had with simple things from Google and others in policing, these algorithms can be wildly inaccurate. It doesn’t take much to throw them off.
Now factor this into it: the speed with which news events move on social media. Many news events can develop their own unique language rules, where certain words take on different connotations because of the context of a news story. This situation can happen rapidly in a developing news environment. Algorithms learn fast, but they’re going to get tripped up by these events.
What I’m saying is that Facebook can’t police speech. There’s no rhyme, reason, or way they can do this. You have to rely on algorithms because humans can’t handle the volume of information or the kind of trauma they’ll encounter. Facebook had to settle for $52 million with a group of human moderators because they developed PTSD from viewing all the insane things people posted.
Not only can’t Facebook patrol speech, but it also can’t do it in a way that benefits one political party. I know the popular thing on the right is that conservatives are hurt more on social media, and there’s some truth to that. Donald Trump is the most prominent example. But algorithms cut both ways. And these policies will impact both sides of the political aisle. Give any algorithm enough time, and it’s going to enrage both sides of the political aisle.
All of which brings to this part of the newsletter: what to do about it???
I recently wrote about how Republicans have started shifting on the issue of antitrust, and we’re going to dig into that here a bit more. If you’ve read me long enough, you know I’m more of a civil libertarian on most topics relating to rights. But this is one topic where my thinking has become more influenced by the politics of the moment, which I believe are becoming harder to escape.
Gaming out the situation.
Here’s what I mean: as I pointed out above, the speech algorithms that Facebook and other social media platforms deploy will curb the speech of Democrats and Republicans. And as those algorithms start hitting politicians, the urge to “do something” will rise on both sides of the partisan divide. Some politicians will inevitably use social media attacks on them for fundraising and increasing their profile, but others will take action.
And the political winds will be with those wanting action. These are the political fundamentals of our moment. If you’re a civil libertarian denying these facts, your head is buried in the sand. The only way to eliminate these impulses in both politicians and the masses is for social media platforms to revert their internal policies to a pre-2016 election stance on social media, where anything goes (except criminal actions).
Are there any takers on that? I don’t think so. I don’t see Silicon Valley ending its belief that it alone, in all its wisdom and wealth, and single-handedly fix the country. From what I can tell listening to them talk to tech-friendly outlets, they fundamentally believe they can make things better by clamping down even more.
If BigTech is not inclined to “lay down their arms” in this fight, then you can’t expect politicians on both sides of the aisle to sit idly by and do nothing. Democrats and Republicans will want a crack at BigTech companies for differing reasons. All this adds up to a higher likelihood of action by one of the parties if they get a majority in Congress.
There are three options as I see them. You can go after tech companies directly or indirectly. By direct, I mean things like new speech regulations on those platforms or altering Section 230. Of all the options available, I consider these the very worst because it introduces new regulations into the US Code, which is not good. Anything new here has questionable constitutionality too.
Antitrust as the answer.
That brings us back to indirect pressure through two avenues: increasing taxation on tech companies or strengthening antitrust laws. The taxation point, especially for the right, is more bark than bite. And all it does is raise the barriers of entry against new tech companies, or BigTech companies can move their headquarters overseas.
That leaves antitrust as the least worst option available for channeling political anger at these companies. And it’s the least bad option because, as Rachel Bovard points out: “Antitrust enforcement is not regulation.” She fleshes that argument out here in the American Conservative. The case against this comes most prominently from the Alliance on Antitrust.
There are solid, principled arguments on both sides. I’m coming at this from the political angle, however. As I said at the top, the political situation here concerns me. What the Alliance on Antitrust and other civil libertarian types don’t seem to understand is this: tech companies will not relent on policing speech that warps the public discourse. The status quo cannot and will not hold.
Democrats and Republicans will unite on one point: something has to change. The broad contours of what people think needs to happen have already been filled out. This kind of political energy is going to get channeled into one of those outlets. You cannot oppose all of them, like William F. Buckley standing athwart history yelling stop, and expect to win. Political momentum is going to carry us over this cliff unless the tech companies relent.
In that frame of mind, the prudent politician should seek to channel those impulses into the least worst option of all. For my money, that best bet is antitrust. It hits these companies where it matters to them and doesn’t involve new regulation.
Antitrust is a way to channel this political and populist anger at the companies themselves without creating new regulations. Theoretically, you could do the same thing by threatening the companies with higher taxes. But they could relocate to another country to avoid taxation. Antitrust is the safer pathway with a chance of channeling the anger, as well as dramatically reducing the cultural power a single company can have in the process.
There are distinct downsides. I don’t deny that. But I also don’t believe you can hope for these impulses to leave Democrats or Republicans. And the benefit of antitrust is this: if it ends up being a bad policy, you haven’t increased government power in the process. Eventually, the free market will fix things, much as we witnessed phone companies create a big three in the form of AT&T, Verizon, and the Sprint and T-Mobile merger.
Economic nature can heal over time. Repealing new regulations might not happen if they’re truly awful.
Ideally, tech companies stop this crusade they have on policing speech. If they give it up and return to a pre-Trump stance, I think things cool down, and everyone is okay. But the more they push this, the more likely action becomes. And if that happens, intelligent politicians have to figure out how to channel that anger without creating new headaches. Antitrust is the best solution.
Links of the week
A Short History of China’s Biohazard Accidents — before COVID-19: With suspicions growing over a possible lab leak tied to the coronavirus pandemic, China’s startling lab-safety record merits close examination. – Jim Geraghty, National Review
The Groupthink That Produced the Lab-Leak Failure Should Scare Liberals – Jonathan Chait, NYMag
The Lab-Leak Theory: Inside the Fight to Uncover COVID-19’s Origins: Throughout 2020, the notion that the novel coronavirus leaked from a lab was off-limits. Those who dared to push for transparency say toxic politics and hidden agendas kept us in the dark. – Katherine Eban, Vanity Fair
NIH Director: We Need an Investigation Into the Wuhan Lab-Leak Theory: Francis Collins calls for a “thorough, expert-driven, and objective” inquiry, and shares what most surprised him about the virus. – Peter Wehner, The Atlantic
She Got Pregnant. His Body Changed Too: Couvade syndrome, in which men get pregnancy symptoms, remains a medical mystery. But the condition reveals how transformative fatherhood is—and how society misunderstands that. – Ariel Ramchandani, The Atlantic
Maggie Haberman Is Right – Donald Trump believes he will be reinstated in August – Charles C. W. Cooke, National Review
America the Outlier: Voter Photo IDs Are the Rule in Europe and Elsewhere – John R. Lott Jr., RealClearInvestigations
Woke Institutions is Just Civil Rights Law: Why Conservatives Won’t (and Can’t) Fight for Influence, and What to Do About It. Followup to “Why is Everything Liberal?” and “2016: The Turning Point” – Richard Hanania
Twitter Thread(s) of the week
Satire of the week
Thanks for reading!