Howdy, picnickers!

You are joining me during a historic (an historic? brits in the audience, give me a shout out) rout of tech stocks, on the back of investor doubts that the historically heavy investment will ever translate into profits.

It would be one thing if the AI industry could point to some huge downstream impacts that these investments have produced. But even AI-friendly Atlassian has acknowledged in their State of Product report that while 80% of teams claim AI is helping them ship faster, the benefit to customers is zero.

And so we must once again probe the thorny topic of: if all this code we are shipping is not making things better for users, what exactly are we doing here?

User needs vs user rights

User-centered design (often reshuffled as customer-centered or human-centered) has been de rigueur for the past two decades. It’s interesting, then, that user rights have only really began to pop up recently.

The user experience is now a matter of law, and the law has some interesting things to say. Foremost among them has been Meta’s repeated legal defeats over its exploitative algorithmic feed. Its deliberately addictive design has been slapped with a $17 billion fine and Meta has been compelled to make a wide range of changes to its platforms’ design aimed at protecting its users. Earlier this year, TikTok was also found to have created an experience that was too addictive for users.

But how did these experiences — in all their behavior-changing splendor — come to exist in the first place? That is the real matter that our industry must reckon with: the best efforts of the highest-paid designers resulted in a crime. And while it’s easy to blame someone else (“the business!”) on this unfortunate outcome, the heart of the matter is inherent to how all design is done nowadays. After all, weren’t all these services once heralded as the paragons of great design? What made them (dare I use the term) enshittify?

The notion that the design of our software is getting worse is not new. The epistemic violence of updates no one asked for was captured in a classic 2012 essay:

Releasing an update is practically an act of aggression against your users. The developer perspective is "You guys are going to love this new update we've been working on!" The user perspective is "Oh god here comes another update, is there any way I can postpone the agony for a few more days?"

— Jono Xia, Everybody hates Firefox updates

Christina Wodtke’s cat-appareled opus is another example of the early 2010s acknowledgement that what we were doing was no longer “user-centered” but something else: a burden we impose upon the real people using our software by unilaterally moving things around. Our touchpoints for what constitutes “good design” have remained unchanged since 1989 because they were finished once they shipped, while our most hated products lose our faith when they continue to mutate beyond our wishes.

The attention economy rules everything around me

One of the invisible constraints of UX design (acknowledged since before it bore that name) was the scarcity of attention: regardless of how many users you might have in aggregate, an individual user only has so much attention to go around. As the always-on push notification based world established itself, that attention grew more and more scarce, but our product metrics have scarcely kept up. Originally, machine learning was employed to help users decide what is worth paying attention to. But the scale of “recommendations” has far exceeded users’ ability to keep up.

In a way, the discipline of content design burst forth, phoenix-like, from the ashes of information architecture precisely because someone needed to deal with the consequences of information scaling beyond the ability to actually use it:

The capability to store and transmit content will forever outpace the capability to structure and interpret it.

— Chappell Ellison

Content design has done a lot for meeting users where they are, pushing back against the path of least resistance created by chatbots. The push towards saying things simply has been a breath of fresh air. Writing for people, not businesses remains one of the key principles of good content. And (as I am always a fan of productive friction) the importance of docs to the UX of ecosystems cannot be overstated:

The cost of good docs was never the writing. It was the friction that forced us to think, the tension that made us consider every reader in the room, and the truth that only comes from caring about the product and the people who use it.

— Fabrizio Ferri Benedetti, The writing was always the cheap part

Your dashboards are lying to you, because they are dashboards

One of my biggest white whales is the influence that trendy form factors have over software products. Of those patterns, the dashboard is perhaps the largest, whitest, and whaliest. Managers in all industries can never shake the idea that a “single pane of glass” can somehow give them the information they need to make better decisions than the people actually involved in doing the work. But those same managers usually over-estimate their own ability to participate in the information economy, and how dialed-in a dashboard must be in order to truly be actionable:

The perfect dashboard has two states. Green, or the name of the person stopping it being green, in red.

— Bluesky post

Any irrelevant information presented on a dashboard constitutes a cognitive burden. You may try to ameliorate it through UI tricks (for example, color coding — did you know that in a factory, red is “on” and green is “off”?) You may even try your hand at data viz ink golf. But the only winning move is not to play: after a certain point, the amount of content we shove at our users can no longer be meaningfully processed.

You can’t AI your way out of information overload

Did you know that AI can summarize your emails? Or even (within a certain tolerance for embarrassing error) respond to them? Several hundred million dollars of advertising budgets have been deployed towards making sure you are aware, so you better have caught on.

Does any of that AI summarization do a goddamn thing for helping anybody process and respond to events? Absolutely not, according to UK Gov studies. While satisfaction levels are high, actual time savings are “extremely small” and quality is below what baseline employees achieve without AI help. In other words, AI models still bow before the iron laws of the attention economy.

Unfortunately, what AI is extremely good at is overloading the attention economy with high volumes of nearly-correct slop:

The speed is very much real for each of us individually. It’s just that when you add it up across the team, it points the wrong way; everybody’s faster, and the whole thing somehow moves slower.

— Matheus Lima, You got faster. Your company didn’t.

Thanks to generous VC subsidies (expiring soon!) the cost of sending an AI-generated letter of concern is near zero. The cost of determining whether the concern is warranted — and responding to it — has not actually gone down at all. The result is an economic imbalance that model providers are happy to pitch as a benefit of their service, even as actual services buckle under the weight of supporting the influx of demand.

Thankfully, today’s designers are paying attention to cognitive accessibility as one of the dimensions of an accessible user experience. As the demands of “the business” add yet more complexity to products, it’s absolutely critical for our field to develop a framework for reconciling those demands with the needs of users who are tired of our shit.

— Pavel at the Product Picnic