2020

I decided to try out this lossless text-compression demonstration site by Fabrice Bellard. It uses GPT-2 natural language generation and prediction to achieve compression. As sample text, I used the first paragraph of Donald Trump’s recent rally speech in Tulsa, Oklahoma. (I figured if anything can compress well using predictive machine learning, surely Trump’s speech patterns can.)

Here’s the compressor site, with most of the input and all of the output showing:

compression page with both input and output displayed

The output looks like a short string of Chinese characters because the compressed text is represented as a series of Unicode characters (encoding 15 bits of information per character — which makes the compression ratio displayed, 804/49, a bit misleading, since the characters on the bottom are twice as large as the characters on the top: 402/49 would be more more accurate, and still quite impressive).

Anyway, I naturally thought “Hmm! What would happen if I were to paste this presumably random Chinese output into Google Translate?”

I am a prisoner, and I am in a state of mind.

“I am a prisoner, and I am in a state of mind.”

Aren’t we all, Internet? Aren’t we all?

Have you noticed how Trump consistently says that “we can’t let the cure be worse than the problem“? (emphasis mine)

The usual stock phrase ends with the word “disease”. But Trump avoids the stock phrase, probably because he doesn’t want someone quoting it back at him sarcastically at the peak of the COVID-19 death toll. So in order to avoid reminding his listeners that it is, in fact, literally a disease we’re dealing with here, he twists a common saying.

Since Trump’s use of language is so frequently odd anyway, journalists rarely call out his misdirections or try to explain them. But even worse, they often cover for him. There was a particularly dramatic example of this recently:

On The Daily podcast with Michael Barbaro, New York Times journalist Maggie Haberman played audio of Trump saying “I don’t want the cure to be worse than the problem itself” (he always phrases it this way — he never says “disease” in that phrase) and then she did a really interesting thing. She repeated it back for the audience, but with the phrase corrected to its standard form:

“— in his words, the cure can’t be worse than the disease.”

(Here’s a transcript.)

Haberman wasn’t adding any information by rephrasing the President. She wasn’t summarizing a longer or more complex thing Trump said. She wasn’t providing needed context that the listener might not have. She just repeated Trump, with one important fix — and called her fixed version “his words”.

What is going on? It’s not a simple accident. The day before, Michael Barbaro himself did the same thing. He played audio of Trump using the same odd phrasing on a different occasion, and then Barbaro followed it up by similarly fixing the President’s words, albeit with “illness” instead of “disease”. (transcript here)

It’s as though the journalists know something is wrong, and instinctively want to fix it, so they generously clean up after the President, instead of simply pointing out how the President consistently mis-phrases a traditional saying. (Foreign journalists have noticed this tendency of American reporters to edit the President and thus mask what he’s actually saying.)

I’m not suggesting that reporters should indulge in speculation about the President’s motivations in behaving like this, even when those motivations are pretty clear. Instead, I’m suggesting that journalists should point out when something odd is going on — help the audience see patterns. As reporters, they’ve heard Trump use this strange phrasing multiple times; they know full well what is going on. But any given audience member might not have heard all those instances, and thus might not spot the pattern.

Instead of unconsciously correcting Trump, and thus normalizing him, just report on him and help people be aware of patterns. Listeners can come to their own conclusions about what the patterns mean, but no one is in a better position than journalists who cover Trump professionally to point out the patterns in the first place.

Don’t cover for.

Just cover.

(Note: See related Twitter threads here and here.)