Birdwatch, week 2: Are You Experienced
The subtleties of the Birdwatch voice are subtle. But the first few rules are simple enough even a computer can understand them.
editorial note: next Wednesday, the Twitter annual shareholder meeting. Hopefully, Elon Musk will ask some of his questions on a live call, rather than this cat-and-mouse game we are currently watching.
Last week, we came to Birdwatch for the first time:
Today, we are back, for a second week of thoughts. This week, we focus on one tweet, and the various ways Birdwatch users interact with it.
the Context Splice
the Newslettr’s editorial staff has substantially more1 experience with Birdwatch than the median new-user. Of course, the median new-user has no experience whatsoever.
Birdwatch responses on this (all slightly edited2) include:
The Birdwatch comments are all stupid. Why does this box even show up?
Elon Musk is stating his opinion. This is not a matter of fact.
Joe Biden is out-of-touch, according to a recent New York Post op-ed.
According to Scientific American, most of the division and hate in America is being caused by Republicans.
Opinion tweets should not be fact-checked.
Let’s start with the first one. “The Birdwatch comments are all stupid. Why does this box even show up?” is certainly a commentary on the concept of Birdwatch. Yet the new3 user didn’t trigger a context-switch.
The map is not the territory. The intent of the user wasn’t “this message should be displayed for all eternity as the Birdwatch commentary for Mr. Musk’s tweet”. The intent was “I have one opinion and there is one text box here”.
“Elon Musk is stating his opinion. This is not a matter of fact.” and “Opinion tweets should not be fact-checked.” - Once again, technically correct, but not the type of thing that should be in a visible commentary box.
If I remember correctly, evidence for “not misleading” tweets aren’t supposed to be displayed anywhere. Perhaps they are displayed in the app somewhere If so, they are fodder for “Note not needed on this Tweet” feedback.
There is an interesting design choice here. If a tweet isn’t misleading, that is an important signal. But the “this is accurate commentary about a tweet” and “this should be displayed with a tweet” wires get crossed.
the Newslettr suggestion: a “not misleading” Birdwatch comment should be displayed only when people try to submit a Birdwatch comment. There needs to be a be a visual indicator4 that you are rating comments for a separate purpose.
There is one additional context-splice that has come up a lot. People see the “Help rate notes submitted by Birdwatch contributors” interstitial, and interpret that as “Twitter feels this Tweet is misleading or inaccurate”.
The first idea that comes to mind is some form of a “temperature gauge” for a tweet. A photo of an ice-cream treat from a Disney resort would be very cold, a comment about the motivation behind a recent felony would be very hot.
The Worst Kind of Correct
“Joe Biden is out-of-touch, according to a recent New York Post op-ed.”
First: I don’t want to ever hear about the New York Post5. And an op-ed is little more than a fancy blog post.
Second: This has nothing to do with Elon Musk’s tweet. Somebody is just putting their own opinion in the wrong box again. This shouldn’t be Birdwatch, it should just be a regular tweet.
Third: This is off-topic. It is in the same category, I suppose: 2020s American Politics.
“According to Scientific American, most of the division and hate in America is being caused by Republicans.”
First, if you check the link ( https://www.scientificamerican.com/article/why-hatred-and-othering-of-political-foes-has-spiked-to-extreme-levels/ ), you will find the summary provided isn’t particularly accurate. Also, it is from October 2020.
Second, after years of experience, the fact that a scientific study has come up with these conclusions is evidence of exactly nothing6. The choice of which scientific studies get covered in major news outlets is more a statement about the ideology than the scientific value of a study.
Third, presumably Mr. Musk has more than 280 characters of thoughts on the topic. It might be an interesting thesis for a 20-minute debate. As a 280 character comment, it is just a cheap-shot (or a non-sequitur).
The Great Grey Hope
Last week, I commented that while “GPT-3 writes Birdwatch comments” would be nice, it isn’t going to happen this year.
The comment reviews, on the other hand, might be able to be done by AI. At least the context-splices and the off-topic opinions can be filtered out.
Staff resumes are not available for newslettr.com contributors.
Either for copyright reasons, or for privacy reasons. Take your pick.
It is safe to assume that every user is a new user at this point in time.
the Newslettr’s advice sometimes feels like a magic 8-ball. In this case, “have you tried having a visible icon to specify context”.
This is one of those stories where you actually have to buy us a drink in person first. You don’t have to drink the alcohol, but you do have to pay money for it.
If you don’t know what the file-drawer effect is, you can stop reading now. If you don’t know what p-hacking is, you can stop reading now. If you don’t know what cherry-picking is, you can stop reading now.
Amazingly, the ruckus has quieted down.