Too often, a popular consumer app unexpectedly gets worse: Some combination of harder to use, missing features, and slower. At a time in history where software is significantly eating the world, this is nonsensical. It’s also damaging to the lives of the people who depend on these products.

First, a few examples to clarify the kind of thing I’m talking about. These are just the ones I’ve had personal experience with.

iPain · One super-obvious example is the long, sad story of iPhoto and iMovie.

For years after the introduction of iMovie ’08, you could still get and use the ’06 version and lots of people did, because it was simple, straightforward, and the obvious things you needed to do were always within reach. I was using the program back then and since I’m a tech geek updated to the newest and greatest, then was reduced to inchoate screams of rage by ’08. I couldn’t figure out how to do lots of obvious things, everything was klunky. There wasn’t a single dimension along which ’08 was better.

As for iPhoto, I never used it much, but my eighty-something mother did, and took lots of great photos with the Sony RX100 I gave her when I gave up on pocket cams. She’s not geeky but has a Bachelor’s in the sciences and is really smart. At some point they broke iPhoto so she couldn’t figure out how to do anything, and when she asked me for help she had tears in her eyes. I tried to get her fixed up, but she doesn’t take pictures much any more. I miss them.

Economist pain · I was still a Developer Advocate in the Android group when The Economist shipped their app. I thought it had one of the best user experiences ever. You started at the beginning of the current issue, swiped down through an article to the bottom, then swapped to bring the next article in from the right. It remembered where you’d got to, which supports The Economist’s vision of being a weekly newspaper; one pass through and you’re caught up on the world that week. There was always a gesture to get to the Table of Contents, but I found I usually didn’t need it much, just swiped over the things I didn’t care about. I praised it to the skies at the time, and (admittedly) since criticized its “Back” affordance, but that was a minor gripe.

The most recent version has been fancified and crippled. First of all, when you open the app, it doesn’t take you to where you were last reading. It insists on starting with “news of the day” (there are lots of other sites for that stuff) and you have to press “week” to get back into the actual publication. When you do that, even though it knows which articles you’ve read (marking them with a check-mark in the table of contents) it maddeningly doesn’t take you to where you were last. So you have to hunt through the table of contents to get yourself restarted.

And when you get to the bottom of an article, it doesn’t stop, it drops you into some weird bastardized section-specific table of contents thingie. All I want is to flip down then flip right until I get to the damn end of the damn magazine. Why?!

MLB · I’ve used the Roku/MLB combo for years to watch ball games on our big TV. The app has evolved over the years and mostly gotten better. I find things on Roku tend to be a little sluggish, but MLB wasn’t too bad; it’d drop you, pretty quickly, into a screen containing a nice picture of a baseball stadium, then overlay a grid of games that were on; pick the one you want and away you go.

Suddenly, it’s become immensely slower, and apparently is spending that time trying to use some AI voodoo to figure out which game I’d like to watch. After an endless delay, you get live video of the game it thinks you want to watch, with a few other games and menu choices overlaid around the edge. It’s reasonably good at guessing which game I want to watch, but way slower at getting me there than it used to be. When’s it’s (regularly) wrong, there are two (slow) menu transitions to get back to the grid of all the games.

Also, they screwed up the Android Auto app — I find listening to a game a good way to pass the time on the road. It’s always had a flaw in that it tries to guess which game you want to watch and starts playing that — the guesses are laughably bad and I often end up with something like Miami Marlins in Spanish. But, you were one tap away from a nice list of everything on offer.

Recently, the startup screen is trying be smarter, thus much slower, in presenting its guess as to what you might want to hear, with a few others (not all) offered as options. So I have to wait forever for this to manifest then hit a teeny little “More…” target to get the actual list of all the games on offer.

Why does this happen? · It’s obvious. Every high-tech company has people called “Product Managers” (PMs) whose job it is to work with customers and management and engineers to define what products should do. No PM in history has ever said “This seems to be working pretty well, let’s leave it the way it is.” Because that’s not bold. That’s not visionary. That doesn’t get you promoted.

It is the dream of every PM to come up with a bold UX innovation that gets praise, and many believe the gospel that the software is better at figuring out what the customer wants than the customer is. And you get extra points these days for using ML.

Also, any time you make any change to a popular product, you’ve imposed a retraining cost on its users. Unfortunately, in their evaluations, PMs consider the cost of customer retraining time to be zero.

How to fix this? Well, in my days at Amazon Web Services, I saw exactly zero instances of major service releases that, in the opinion of customers, crippled or broke the product. I’m not going to claim that our UX was generally excellent because it wasn’t; the fact that most users were geeks let us somewhat off the hook.

Why no breakage? Because these were Enterprise products, so the number of customers was orders of magnitude smaller than iAnything, so the PM could go talk to them and bounce improvement ideas off them. Customers are pretty good at spotting UX goofs in the making.

The evidence suggests that for mass-market products used by on the order of 10⁷ people, it’s really difficult to predict which changes will be experienced as stupid, broken, and insulting.

Maybe we ought to start promoting PMs who are willing to stand pat for an occasional release or three. Maybe we ought to fire all the consumer-product PMs. Maybe we ought to start including realistic customer-retraining-cost estimates in our product planning process.

We need to stop breaking the software people use. Everyone deserves better.