I don't pick a restaurant anymore without checking what strangers thought of it first. I don't hear a song until something has already calculated that I'll probably like it. I don't drive anywhere without a route that has quietly removed the wrong turn before I had the chance to take it, and I don't meet many people without having seen a profile first.

None of that is bad. That's what makes it worth sitting with for a minute. Every one of those tools works, which is exactly why I use them, and why I'm not going to pretend I'd be better off throwing them out.

But somewhere in there a habit set in that I didn't choose on purpose. I've started requiring evidence that something will be worthwhile before I'll let myself go have it.

That's the part worth being precise about, because the overstated version of this argument is wrong. Chance hasn't been eliminated. A recommendation engine can genuinely hand you unfamiliar music. Maps can get you to a place you'd never have found on your own. The shift isn't that surprise became impossible. It's that I started asking for proof first, and proof is only ever available for things that already have a track record.

What Rankings Would Have Cost Me

Here's the one I think about most, because if I'd followed the evidence I would have missed my entire life.

I was deciding on a master's in finance. I had admits from Berkeley, from Michigan, and from Indiana, and Indiana's Kelley School was the highest ranked of the three. So I accepted it. That was the correct decision by every available metric, made the way you're supposed to make it, which is by looking at the rankings and picking the top of the list.

And then something, and I'll say plainly that I believe it was God, kept telling me to apply to Georgetown. The deadline had already passed. I had already accepted somewhere else. On paper there was no argument for it. I applied anyway, late, with no good reason I could have articulated to anyone who asked.

They admitted me.

I met my wife there.

Think about what the defensible process would have produced. The rankings were real, the timeline was real, and following both would have been the responsible thing by any standard I could have defended to a reasonable person. I'd have gone to Kelley, which is an excellent school, and I would have gotten a good education and no idea whatsoever what I'd missed. That's the quiet part about optimizing on evidence. When it costs you something, it never sends a bill. You just don't find out.

The degree itself I was able to get because of the GI Bill rather than six figures of debt, which is its own piece of luck I had nothing to do with. But which building I was sitting in when I got it came down to ignoring every number in front of me.

The Smaller Version

It happens at a much lower stakes level too, and the mechanism is identical.

A couple of years ago I was bored and genuinely annoyed with Fairfax. No plan. I just drove around near where I used to work in Arlington, which is the kind of aimless thing that has no justification and produces nothing you could have forecast. And I found a building that was brand new and mostly empty, which meant they were giving away two months free at a lease price that will never exist again once a building fills up. Top floor. I lived there for two years and loved it, and I flooded the place with memories.

Not one step of that survives a search filter. The building had no reviews because nobody lived there yet. It had no ratings, no history, nothing anybody could have vouched for. Its entire advantage was that it was new and empty, which is precisely the profile that a system built on accumulated evidence is structurally unable to recommend. The only way to find it was to be physically present, with no agenda, slightly irritated, driving around.

This Isn't the Algorithm's Fault

I've written about how recommendation engines hand you a mirror instead of a window, narrowing what you see down to a sharper version of what you already are. That's a real thing and it's a different thing from this.

This one is mine. The feed didn't stop me from applying to Georgetown. No algorithm kept me from driving to Arlington. What almost stopped me both times was my own requirement that a thing demonstrate its value before I'd commit any part of myself to it. The rankings didn't impose themselves on me. I went and got them, because I wanted something external to tell me the decision was defensible.

That's the part I can actually do something about, which is why it's the more useful problem of the two.

And it connects to something I got into recently about how the question of what to be interested in quietly stops being yours. This is the same move one layer down. First you let something else decide what's worth your attention. Then you require it to certify the experience before you'll have it.

Where the Habit Is Right

I want to draw the line carefully, because there's a version of this that gets people hurt.

Some uncertainty is consequential and asymmetric. Check whether the contractor is licensed. Read the terms on the loan, especially the ones written so that one slip pays for everything. Verify the person you're meeting is real. Those aren't small tolerable unknowns, they're the kind where being wrong costs you years, and the tools that reduce them are a genuine improvement over how our grandparents had to live.

The subject here is the other category. The small, cheap, recoverable uncertainties. A mediocre dinner. An hour in a neighborhood that turned out boring. A record you didn't like. A drive that went nowhere. Those have a downside measured in a wasted evening, and I've been treating them like they deserve the same verification I'd give a mortgage.

The cost of a bad restaurant is one bad meal. The cost of never going anywhere unverified compounds for years, and it compounds invisibly, which is the only reason we tolerate it.

What the Two Stories Have in Common

In both of mine, the incomplete information wasn't a cost I paid. It was the condition that made the thing available at all.

Georgetown was only on the table because I was willing to act without a reason I could defend. The apartment was only available because it had no track record yet. In both cases the absence of evidence wasn't a risk I accepted reluctantly. It was the whole reason there was an opening, because anything with enough evidence behind it to be certified has already been found by everyone else who was waiting for certification.

That's the trade nobody describes when they hand you a better filter. The filter does genuinely protect you from the bad version of the experience. It also quietly restricts you to the experiences that have already been had enough times to be rated, and the thing that would have changed your life has, by definition, not happened to you yet.

So I've been trying to keep a few decisions unverified on purpose. Not the ones that matter in the way a loan matters. The small ones, the evenings and the drives and the unproven places, where the worst case is that I waste a couple of hours and the best case is that I meet my wife.

A life with no accidents is easier to manage. It's just harder to remember.