Forward momentum creates stability.

      

That’s the wrong question to consider when proposing a study.

It’s also not helpful to say, “It’s unlikely to solve the problem.”

All the likely approaches have already been tried.

The useful steps are:

1. Is there a problem worth solving?
2. Is the expense of this test reasonable?
3. Will the study cause significant damage?
4. Of all the things we can test, is this a sensible one to try next?

Our fear of failure is real. It’s often so significant that we’d rather live with a problem than face the possibility that our new approach might be wrong.

If the problem is worth solving, it’s probably worth the effort and risk that the next unproven test will require.

[In this podcast, Dr. Jonathan Sackner-Bernstein talks with some patients and a doctor about his novel approach to Parkinson’s disease. Participants in the conversation bring up the conventional wisdom he’s challenging and share reasons why his theory probably won’t work. But none of the critics has a better alternative. The cost of the test is relatively low, and the stakes of the problem are quite high. There’s no clear answer. This is precisely what a study is for.]

What will it cost to test your solution to our problem? Okay, begin.

      

Imagination is a skill and it takes effort.

It’s not useful to say, “I’m not imaginative.”

It’s more accurate to realize that we might not care enough to get good at it, or to put in the effort it takes.

As tasks continue to be automated, the hard work of imagination is worth investing effort in.

      

18 years ago, I posted this image:

…and I still can’t get it out of my head. Sorry.

Why do we have such a creeped-out reaction to images that aren’t quite right? A robot that looks too much like a person, or a song that we can somehow tell has an AI voice.

The creepiness predates AI, and was first named in a paper by Mori fifty years ago. But it’s so visceral that it almost certainly originated along with our fear of snakes and other evolutionary safeguards.

There are probably two things going on.

First, there’s a corpse alert. Corpses are dangerous, and something that’s alive/not alive is a warning sign. Same thing with zombies.

Second, imposter alert. Imposters are even more dangerous than predators, and we honed our imposter-detection skills a long time ago.

And now, everyone has AI available to them, and many of us are churning out experiences that border on the uncanny valley.

Not many people care about an automated drum track on a pop single, but we get uncomfortable when the lead singer isn’t quite human. We don’t mind when a website figures out our zip code for us, but when a bot apologizes for a late shipment, it means less than nothing. We’re okay with animation, but not with an educational video that combines beautifully shot real footage with an animated human that’s almost but not quite real…

While it’s possible to get used to snakes, and, perhaps, to corpses, I’m not sure that the general population is in any hurry to get used to either, or to the uncanny valley.

It’s likely that AI quality will increase fast enough that many of the most egregious valley moments will stop happening. But none of that will help with the expectation chasm. When you install an AI admin, or use AI for customer service or therapy, we will always end up with a valley sooner or later.

The solution is simple but takes effort: don’t fake it. Celebrate your genre, make a promise and keep it. Not in the way we need to label the ingredients in food, but simply to avoid the surprise realization, to protect your customers from the ick. Triggering an evolutionary survival mechanism is rarely good for your career.

“I confused and alienated people as I worked to save money trying to get them to think this was a person” is not much of a mission statement. Our job is to find problems and solve them, not to hustle our way with shortcuts that feel creepy.

Three videos for today:

The talking dog and AI.

Hank Green on the essential Mola sunfish metaphor.

Talking with Jon and Becky about We are For Good and the work of non-profits.

      

“I can’t imagine eating durian ice cream,” is not the same as “no one likes durian ice cream.”

We fail as marketers, editors and project managers when we can’t find the empathy to bridge the gap. It’s a lovely shortcut to make things for yourself, to imagine that you are the client, the reader or the customer. But most of the time, you’re not.

“It’s not for me, but it might be for you.”

      

Setting out to make something popular requires only a focus on the crowd and on the moment. Most pop music is popular simply because that’s what it was built to do.

Good work can be good without being popular. And so the two goals aren’t easily aligned.

It helps to begin by becoming comfortable with what good feels like to you. Because conflating it with popular is a trap.

      

There’s a difference between casual online games that have a right answer, and those that are open-ended.

In crossword puzzles and most of the games from the Times (like Wordle and Connections) you’re trying to guess what the puzzle constructor had in mind. This can lead to frustration, because the idiosyncratic nature of inventing clues and answers means that you might not be in sync with the person at the other end. They’re inherently closed systems.

Bongo, on the other hand, is generative and combinatorial. There are bazillions of possible right answers, and your goal is to find a right answer that’s worth more points than anyone else’s. It doesn’t matter that I invented the game, I have no advantage over everyone else, because we all begin with the same tiles.

For me, open-ended games are time well spent. Have fun.

      

I just read the first two pages of a sci-fi novel on my Kindle. The author proudly proclaims that the 400-page book was created without any AI whatsoever. Alas, the book is slop. The writing is overwrought and the dialogue is banal. If a page isn’t worth writing, it’s unlikely a chapter is.

Slop happens when a marketer who should know better stops trying. It’s when we prioritize volume over impact. If we measure the cost of what we create instead of its value, it’s likely we’ll end up with slop.

AI makes this easier, no doubt. But it pays to focus on avoiding slop, not in worrying how the slop is made.

The question is now, “Who approved this?” not “who made this?”

      

Actually, they won. They didn’t just win because they got a nearly $3 billion breakup fee.

They won because in just about every contentious public auction, the winner is the one who is willing to overpay the most.

The best way to win a bidding war is to not bid.

      

I think it’s more productive to imagine that “Justice is how society deals with luck.”

Good and bad.