May 11, 2021

Some Thoughts On Twitter Pulling The Plug On Trump’s Account

By Paul Levy

First of all, corporations
enjoy First Amendment protections, among other constitutional
protections. Although some of my friends decry that proposition,
given the Supreme Court’s current composition, that is not
going to change during my lifetime. And the First Amendment protects
the right to refuse to associate with speech one does not like.
There is only so much that legislation could do to prevent companies
like Twitter from controlling the speech that they allow.

Second, withdrawing or
adapting the section 230 liability shield
is one way to impose limits on platform’s adoption or
implementation of their content control policies, perhaps, but there
is no reason to think that any withdrawal that is likely to pass, and
that would be constitutional (because it does not involve viewpoint
discrimination), would be better than the current state of affairs.
Moreover, that would be a very blunt instrument that could not easily
be calibrated. I strongly support the principles of section 230,
which allow online platforms to decide what speech they will allow on
their platforms by protecting them against liability for speech that
they carry (with very limited exceptions). They are not common
carriers, like USPS or the PSTN’s. (Thus, Apple and Google
could cabin Parler by threatening to deny it access to the App and
Play Stores, and Amazon could deny Parler web hosting services, all
on the ground that Parler failed to successfully enforce rules
against advocacy of political violence. I find it mind-boggling that
people who call themselves “conservative” are railing
about the plug being pulled on a platform for the stated reason that
it allegedly fails to block calls for political violence). And they
are not government bodies, which are (largely) forbidden to engage in
content discrimination, and especially viewpoint discrimination, in
allowing or suppressing speech. Section 230, both as a legal
principle but also as a social principle, not only allows platforms
to tolerate speech that I find abhorrent, but also allows them to
exclude speech that I detest, or speech that I adore.

Third, the drafters of section
230 recognized that platforms would have content policies, and
considered it desirable
for them to have content policies. Indeed, you can’t run a
platform without content policies. Anybody who has ever tried to
moderate a discussion group will fully appreciate these
considerations. And content moderation is HARD. The periodic “COMO”
sessions addressing “Content Moderation at Scale”
explored the difficulties through a series of hypotheticals. What
became clear is that people of good will, even those with relatively
common policy perspectives, trying to apply even the best of content
policies, will get it wrong sometimes – and that is even if
they have lots of time to evaluate a single statement.

It is inevitable that
different platforms will take different stances about what speech
they will tolerate and what speech they will exclude—in part
because of the audiences at which they aim.

The Reasons for Banning

Turning to Twitter, and to
Trump in particular: Twitter has for many years had a variety of
rules about speech that may and may not be posted to Twitter, which
they call their trust and safety guidelines. This includes a ban on
speech glorifying or promoting violence; it also includes a ban on
false statements about the election. (Many years ago I was asked if
I would be willing to be on their outside trust and safety council; I
decided it would not be appropriate for me to do that for a number of

I have never been a fan of
social media companies trying to assess the truth or falsity of
factual statements, or the hurtfulness of opinions and rhetoric
deployed there. The assumption of such a role is likely to lead to
the suppression of voices that criticize the rich and powerful (that
is, those who can afford to hire lawyers to file baseless lawsuits,
or who can deploy government power to attack their critics), or on
abusive law enforcement officials themselves. But Trump’s
misuse of Twitter and other platforms to foment a violent attack on
democratic elections has taken matters to another level.

In the weeks leading up to the
attack by a mob of Trump supporters on the Congress, Trump used both
tweets and other public communications to try to steal the election
from his victorious opponents. At the same time, he tried to use his
connections with officials in several states where he had lost to
induce them to overturn his losses in those states. Then, with his
campaign of frivolous litigation and threats to state election
officials having come to naught, he and his mouthpieces escalated
heir attack on democratic elections, encouraging his supporters to
engage in a physical confrontation with members of Congress in an
apparent effort to delay the certification of his electoral college
loss. There is every reason to believe that many of the supporters
who tried to storm the Capitol believed that they were acting at
Trump’s behest,
and some analyses suggest
that Trump’s language was carefully calculated to encourage the
supporters to use violence. My reading of the language leads me to
conclude that the calls for violence were insufficiently expressed to
make them indictable under the standards set forth in Brandenburg
v. Ohio, 395 U.S.
444 (1969),
just as my view is that his threats to Georgia state election
officials, on a recording,
fell short of criminal threats or incitement.

Assuming, as I do, Trump
cannot be held legally responsible for inciting the riot, he was
still morally responsible for it. And even assuming, as I do, that
he cannot be convicted for these communications, that doesn’t
mean that a private company, unbound by First Amendment standards,
would be unjustified in deeming them a gross violation of its
policies against advocating or glorifying violence, and hence
treating them as a proper basis for denying him a platform.
Platforms are expected to withdraw the accounts of members of the
public who repeatedly abuse their access to infringe
copyright—indeed, the DMCA gives them a powerful legal
incentive to do so. (In my legal practice, I have had to deal with
web hosting services that were considering removal of
consumer-friendly platforms that were receiving bogus claims othat my
clients were hosting defamation or other tortious speech. Despite
their section 230 protection, they often simply did not want to deal
with the trouble).

In that context, it is no
surprise that a platform cancelled the account of a politician who
repeatedly abused his access to foment a riot. And Trump, after all,
is wealthy enough, and remains powerful enough, to use other means to
amplify his speech. Indeed, he has one mainstream news channel and
several smaller ones that are dying to carry his speech.

But, at the same time, Twitter
has allowed exceptions to those rules, and one very prominent
exception is that senior government officials, particularly heads of
state, are given more leeway on the theory that there is public
interest in letting the public know what such people are saying. As a
practical matter, Trump has had almost complete immunity from the
sorts of restrictions that constrained other Twitter users. Much of
what he has posted was a gross violation of Twitter’s rules,
but he was allowed to get away with it. Note that Trump’s hold
on the special exception is due to expire on January 20.

What I think has really
happened is that, for a variety of reasons, Twitter decided to take
away Trump’s special exception, and his ban — based on a
range of past conduct — was the inevitable result of that
change of position. Twitter has both taken away the special
exception a few days early, and made its decision retroactive.
Considering the way in which Trump managed to use his bully pulpit to
incite a violent attack on Congress that was aimed at overturning a
democratic election in which he was defeated, that seems to me to be
within the range of understandable reaction.

Twitter’s Obfuscation

But the reasons
that Twitter gave
for its decision strike me as laughable – recall he was
suspended for 24 hours, then allowed back on the condition that he
delete certain tweets and stop violating the rules. He did delete
the tweets in question, and to my mind nothing he did after being
reinstated violated their rules. He gave appearance of trying to
satisfy them.

Twitter’s blog post
explaining the Trump ban,
asserted that two Trump tweets violated their rules against
glorifying violence. But the two posts they quoted did nothing to
“glorify” violence” What this comes down to is that
Twitter says Trump has been banned because some of his supporters (in
unspecified instances) are allegedly reading his post-suspension
tweets in various dangerous ways. And misreading
what he said, I might add.

One of the tweets praised his
supporters — the 75,000,000 voters who supported him. He
called them patriots. He said they should be respected and should
continue to have a loud voice. That does not encourage violence.

The other tweet said he won’t
be at the inauguration. Yes, a break with tradition, but maybe the
best response is, good riddance!

Twitter says (and some other
reports have echoed these concerns)
that there are plans for armed protests and another attack on the
Capitol. That
is of great concern. But Twitter does not say that Trump is involved
in that planning or that he tweeted anything about them. I did notice
in passing a report that, after his initial 12-hour suspension was
lifted, Trump had retweeted some of those statements. But the
report also said that Twitter had cited those retweets in its
decision and plainly it has not, so the fact-checking of the report
is suspect. I have not been able to locate the report. And, because
the Twitter account has been deleted in its entirety, I can’t
verify the report (and I have not been able to find any screenshots).

Now, when Twitter justifies
its decisions by relying on tweets that do not, in actuality, violate
its rules, it just tends to suggest that what it has done is
arbitrary. And that is not useful.

What May Really Be the

It appears to me that
Twitter’s official views on Trump’s status evolved very
quickly in the past week. Although top executives felt that so long
as he was president he should continue to enjoy his special
exception, Twitter staff apparently were very much of a different
opinion, and forcefully so. There was apparently a staff petition,
and then a large intra-staff meeting, in which Twitter’s top
executives were raked over the coals by their staff for their
inaction against the Trump account.
That may well have mattered.

Additionally, Twitter was
facing intense pressure on Capitol Hill and in the public arena to be
more forceful about Trump’s incitement of the attach on the
Capitol, and I believe they were genuinely concerned that, left with
his Twitter account, he might well have used it to incite further
violence on January 17 (Q being the 17th letter of the alphabet) and
January 20. Not because the two tweets did that, but because the guy
is out of control. The explanatory blog post refers to the covert
planning for January 17 and January 20 repeat attacks, and I think it
is quite possible that Twitter was worried that Trump might abuse his
privileges. I wish the company had just said that (comparable to
Facebook’s explanation) and said that, in
retrospect, they had decided that its initial sanction for Trump’s
previous violations of its rules was not sufficiently severe.

Perhaps More Cynical

First: Both Twitter and
Facebook have cut back on Trump knowing that he is not going to have
the powers of the presidency much longer, and that, indeed, both the
White House and, soon, both Houses of Congress are going to be in
Democratic hands. Just as they went out of their way to propitiate
conservatives who claim (falsely) that social media companies
discriminate disproportionately against conservatives, while those
conservatives ran the Senate and the White House,
these companies don’t want to be adverse to the new power in

Second, and this is a related
point: Trump liked to talk about how much benefit he derived from
his Twitter account, but the converse is also true: Twitter has
profited enormously from Trump’s account, which creates
enormous controversy and hence draws many eyes to Twitter where they
will see ads. A number of people in the tech sector have been saying
that the situation has simply evolved to the point where the benefits
that Twitter was getting from hosting @realDonaldTrump were getting
to be greatly exceeded by the costs.

Third: One exception to
section 230 immunity is for speech that violates the federal criminal
laws. Some people have suggested that federal law enforcement
officials may have reached out to Twitter to warn that if its
facilities are used to incite more riots in Washington DC, such as on
January 17 and January 20, it might face grand jury scrutiny. Now,
to my mind the First Amendment’s Brandenburg
standard would likely bar prosecution for mere passive hosting of
prosecutable incitement; the Brandenburg
standard requires not just incitement of imminent lawless conduct,
but intent
to incite imminent lawless conduct, and passive hosting of speech of
which the host is not aware does not involve intent. But the
possible exposure without section 230 immunity, and needing to rely
only on the First Amendment, might well have been a chastening

Paul Alan Levy is a free speech litigator in Washington DC