Drew Broussard

okay I have a book to write and the next several weeks are both busy (personally) and likely nerve-shredding (geopolitically) so I'm going to *try* to log out of here and not come back for a bit unless something really cool happens. Like, The Party When He Dies level of cool. So text me if, I guess!

you can really tell how much the ground has shifted when suddenly Jonathan Chait is trying to "well, actually" Mamdani into being a secret centrist in the pages of The Atlantic lollll

I felt the Anthropic settlement screwed over far too many authors. And the more that's revealed about the settlement, the more I'm certain we were indeed screwed.

bailed out of three books in a row over the last few days and *sucks teeth* I don't like this feeling

My hypothesis is that people are generally more predictable than they themselves would like to be, but less predictable than businesses would like them to be, and a lot of modern business is about building guiderails to increase that predictability ("personalisation" a big part of this)

"Make me virtuous, but not yet" is one of the all timer great insights, and one thing servives can design for is to present you with a sequence of micro choices, encouraging you away from planning, as that locks you into your basic preferences more which makes you a more reliable income stream

sometimes I just wanna be listening to "Jump" on CD while playing Ski Free on the desktop in the computer room, you know?

you love to see it, I want these fuckers losing sleep over what's going to happen when their pathetic election interference efforts fail to make a dent in november

Would you like a story or story-shaped-object in the mail from me at your birthday in 2027? Because I would like to send you one!

this story! (truly, every story in this issue, but THIS STORY!)

I love this prize — and this shirt!!

Many of those people are among the most poorly paid at a publication. They tend to be there for the learning experience that comes with reading slush. You don't get that from reading slop. Minimizing (we'll never eliminate it) their exposure is good. (Reading that stuff isn't healthy.)

Mistakes and poor judgment in this area will happen whether or not an editor uses a detector. A detector can still be helpful though, much in the same way my process treats filtering as a pressure valve. A secondary benefit is reducing the amount of slop an editor's team is exposed to.

I fully expect my colleagues to continue to mistakes, both in what they reject and what they accept. Editors need to stop making private and public callouts or statements about authors. That just makes any mistakes far worse, puts themselves in legal jeopardy, and amplifies fear. No one needs that.

I outlined what we had to do here: neil-clarke.com/how-ai-submi... It doesn't involve commercial tools or even text analysis, but it still helps. Is it less accurate than some detectors? Almost certainly, but it works well enough to allow us to wait while the community struggles with this issue.

That said, LLMs have increased submission volume and broken trust in a way this field has never seen before. Some places worse than others, but they are all on the same trajectory. If the field intends for open submissions to remain a thing, it has to find ways to adjust or things/people will break.

A detector should NEVER be the final arbiter. It shouldn't be more than a never-deleting spam filter in the process. The editor still needs to review all of those submissions because, like a spam filter, it will never be perfect. This is not a responsibility that can be delegated to an algorithm.

From what I've seen, most editors (and readers) are far from prepared for this. The rather public instances of retractions likely could have been flagged early if detectors were in use, but those editors still wouldn't have been prepared to make the final assessment. That's the big problem.

Honestly, these tools could probably make my life easier. My workload has been increased 25-30% by slop and partial-slop submissions. I've been on the front lines of the impact of this for a while, which has left me better prepared than most and probably in a better position to benefit.

One of those questions is whether or not they would sign a contract stating they would not retain (or train from) submitted for analysis. Pangram was the first (and remains one of the few) that said yes. I'll always trust a contract more than a statement on someone's website.

It's all the more reason to do your homework and research products more deeply. I've been monitoring these companies for years now and have done extensive testing with ethically-sourced works. I even have a set of questions I've asked the companies as a sort of ethical/viability checklist.

A lot of people like to lump all AI into one bucket. It's not helpful. Same goes for all tech companies. Some care. Some steal. Some have betrayed us. Some have not. Trust is hard when "AI" or AI is in the picture. Detection companies aren't equal or immune. The worst of them also make humanizers.

A really good thread, and clarifying, about AI detection and what it actually takes to wade through submissions / new writing in an age of LLM-generated distrust and disinformation.

Today might be rough, so take this abrupt change of destiny.

This is a great ad and Scott Colom is within the margin of error — I know it's hard to imagine, but imagine! And, maybe just maybe, remind the DNC to actually contest seats like this, because sometimes you might be surprised.

Mindbending development from @ftrain.com, whose invention, launched yesterday, will I think take the internet by storm: whole uncrawlable documents that live entirely in a URL. No server, no login, no account (!) Talked with him about it for just a moment: flaminghydra.com/new-super-pr...

Saw THE DEBUT last night at the Woodstock Film Festival and allllll my theater people need to make every effort to see this movie as soon as it is humanly possible. Spectacular, in every way.

will try to answer these in a piece next week but for now i will say that part of the *point* of court expansion, in my mind, is to create a disequilibrium — a potential race to the bottom — that will force a truce.

the kind of leadership the theater needs! be bold! be loud! be open! but don't forget to have some fucking fun too!

I will always and forever be angry that they saw everything opened up by The Last Jedi, including a systemic understanding of the conflict in the universe and slammed it shut because a bunch of whiny perpetual adolescents pitched fits on social media.

We listened to those losers after Appomattox. And after Watergate. And the Bush presidency. And to some degree after Trump I. Those failures helped get us here. We have to drown them out this time.

Next month & again in 2029, a lot of the same pundits, dem leaders, & cultural & corporate elites who said in 2025 that the people had spoken & we needed to bend the knee will say it’s time to move on, look forward not back, to not waste time on investigations. Our job is to tell them where to go.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!1

No joke. I think this is the most important video I have ever seen about mobile games. This guy breaks down how this game psychologically profiles you, and makes the game harder or easier in order to show you ads