Mycelium

The Rest

When Things Go Wrong

When Print Became Delete

Last week, someone I know lost dozens of emails in one morning. Not from a crash or a hack. From muscle memory.

Her email client updated overnight. When she opened it the next day, everything looked fine. Same layout, same colors, roughly the same buttons. She went through her morning routine — check email, scan subject lines, click the button she clicked every day to print important messages.

Except the Print button wasn't there anymore. In its exact position — same spot, same size, same visual weight — was now the Delete button.

Her hand knew the location before her eyes caught up. By the time she noticed the icon had changed, she'd already clicked it forty-seven times.

What Really Changed?

The interface hadn't gotten worse in any measurable way. The new Delete button was actually more accessible than before — closer to where users needed it, according to the redesign rationale. The Print button moved to an overflow menu, which made sense from a usage-analytics perspective: most users rarely print emails anymore.

This was, by conventional metrics, an improvement.

But something fundamental had broken.

Users Don't Read Interfaces. They Remember Locations.

When you use an app every day, the labels start to disappear. You stop reading "Print" and "Delete" and "Send." Instead, you build a mental map of where things live.

Top-left is "go back." Top-right is "my account." Bottom-right in dialogs means "yes, proceed." The button three inches from the left edge, halfway down the screen, means "print this email."

Not "the print button." Just: that location = that action.

I once watched an accountant work in her decades-old accounting software with startling speed. Menus, forms, shortcuts, all executed without looking. Then IT reorganized her desktop icons. Same programs, same names, different positions.

She was completely lost. She hadn't been reading labels for years. She'd been navigating by coordinates.

The Three Systems We Think Are One

When we place a button on screen, we think we're making one choice: "Where should this go?" But we're actually working with three completely different systems at once.

Physical distance. The closer and larger a target is, the faster someone can reach it. This is Fitts's Law: the time to click something increases with distance and decreases with size. Your thumb can tap a large button at the bottom of your phone screen in 200 milliseconds. That same thumb needs 600 milliseconds to reach a small button at the top. When you move the Print button farther away, you're changing the effort cost of that action. Multiply that by forty-seven times a morning, and suddenly the interface feels slower.

Spatial memory. Humans have exceptional spatial memory, even in digital space. Better than our memory for words, for colors, for almost anything else. You know where Gmail's compose button lives. You know where Figma's toolbar sits. You don't read these interfaces anymore — you navigate them spatially, the same way you navigate your own kitchen in the dark. After just a few uses of an interface, your brain starts encoding locations instead of labels. After a few dozen uses, the labels become almost invisible.

Pattern recognition across products. Users don't come to your interface blank. They bring expectations from every other app they've used. Profile icons live top-right because every app puts them there. Destructive actions like Delete are usually visually separated from safe actions like Save. These aren't arbitrary conventions. They're a shared vocabulary refined over decades. When you break them, users feel like they're looking for the bathroom at someone else's house.

The email client broke all three. By putting Delete where Print used to be — in a position that suggested frequency and safety — the interface accidentally said: "This is a routine, harmless action you can execute quickly."

Location Changes Meaning

Location doesn't just affect speed and memory. It changes meaning.

Think about the word "close" in interfaces. "Close" usually lives in the top-right corner. That position has become so associated with "close" that we barely need the X icon anymore. The location is the meaning.

Now imagine if an app put a "Delete Account" button in the top-right corner where "Close" usually lives. Same function as a "Delete Account" button anywhere else. But it would feel wrong. Not because users don't understand what it does. Because the grammar is off.

Proximity creates semantic relationships. Elements that sit close together feel related. When you group "Delete" next to "Save" and "Cancel" with equal visual weight and spacing, you've created a dangerous equivalence. Separate "Delete Forever" with extra space and different visual treatment, and suddenly the danger is clear.

The distance isn't decorative. It's communicative.

When Print became Delete, it wasn't the functionality that broke. It was the conversation between interface and user — a conversation that had been happening through location, through pattern, through accumulated spatial knowledge. And once you break that conversation, it takes a long time to rebuild it. Longer than forty-seven emails.


When "Interested" Means "Rejected"

I opened OkCupid and something felt different. Where there used to be a tab labeled "Likes You" with a small, honest number, there was now "Interested in You" with a promising "99+".

For a moment, it felt good. 99+ people interested in me? But that moment passed quickly when I looked closer.

What Actually Changed

Before: The tab was called "Likes You." As a free user, you'd see an exact count of people who swiped right on you. The top 4 profiles, blurred. Simple, transparent, sometimes painfully honest. If you had 2 likes, you saw "2."

After: The tab is now called "Interested in You." The count says "99+." It includes people who "Liked you" AND people who "Viewed you."

More information, more encouraging numbers, broader definition of engagement. Seems reasonable enough. Until you think about what "viewed" actually means.

What "Viewed" Actually Means

In OkCupid's swipe interface, when a profile appears, you have two options: swipe right (like) or swipe left (pass). There's no skip button. No neutral browsing mode.

So if someone "viewed" your profile but didn't "like" it — they swiped left. They rejected you.

Let's trace the logic:

1. The tab is called "Interested in You" 2. This includes people who "Viewed you" 3. "Viewed you" means: saw your profile and swiped left 4. Swiping left means: not interested 5. Therefore: "Interested in You" includes people who are demonstrably not interested in you

The term "interested" now encompasses its opposite.

The Cost of Misleading Language

The word "interested" carries weight. It suggests intention, preference, choice in your favor. In a dating context, telling someone that a person who rejected them is "interested" isn't a broader definition — it's a lie that feels true until you open it.

OkCupid's motivation is understandable — a bigger number in the "Interested" badge keeps users engaged, keeps them swiping, keeps them paying for premium. But the cost is trust. The moment a user realizes that "interested" includes people demonstrably not interested, every other word in the interface becomes suspect.

Both systems — old and new — are designed to convert free users to paid subscribers. But they do it differently:

The old system was honest but harsh. You saw a low number. You knew exactly what it meant. The value proposition was straightforward: "unlock those 2 profiles."

The new system is encouraging but deceptive. You see a high number. You don't know what it means. The value proposition is manufactured uncertainty: "pay to find out how many actually liked you versus rejected you."

One system made you feel bad with accurate information. The other makes you feel cautiously optimistic with misleading information.

The first creates artificial scarcity. The second creates artificial meaning.

Words should mean what they say. When "Interested" includes rejection, when "Print" lives where "Delete" was — the interface is making promises it can't keep. And users pay the price for broken contracts they never agreed to.