A Forecast That Only Says GO

There is an assumption buried inside most fishing information.

More information should help us catch more fish.

Better forecasts. Better maps. Better run data. Better catch reports. Better weather models. Better cameras. Better communication between anglers.

Put enough of it together and we should be able to eliminate some of the uncertainty between deciding to fish and actually finding fish.

That’s certainly part of what I’m trying to do with the Fishability Index Engine.

But I’ve started thinking there is another side to it.

What happens when the information gets good enough that everybody knows when to go?

Information Creates Pressure

Fishing pressure doesn’t begin at the boat ramp.

It begins with information.

A good report goes up Tuesday.

By Wednesday it’s being shared.

Thursday somebody posts pictures.

Friday the weather forecast improves.

Saturday morning the parking lot is full.

Nobody did anything wrong.

Every angler made a perfectly rational decision based on the information available to them.

I’ve done exactly the same thing.

But collectively, good information can concentrate people remarkably quickly.

The better we become at identifying opportunity, the better we become at converging on it.

That creates a strange problem for something like the Fishability Index Engine.

If its only purpose is finding the highest probability of catching fish, then success is easy to define.

Find the fish.

Identify the conditions.

Send me there.

But I’m not sure that’s enough.

A GO Isn’t Free

When the engine produces a GO, something happens in the real world.

Maybe I drive three hours.

Maybe I launch the boat.

Maybe I fish a particular stretch of river.

Maybe I join twenty other people who reached essentially the same conclusion from the same run numbers, weather forecast and recent catch reports.

One recommendation doesn’t matter much.

Thousands of decisions built from the same information do.

That’s particularly true when fish become predictable.

A migration compresses them into a river.

A tide concentrates them near a point.

Water temperature pushes them into a particular depth.

A run peaks.

A bite gets discovered.

The same information that makes fishing easier can make fish easier to find repeatedly.

And once something becomes predictable, pressure tends to follow.

The Problem With a Forecast That Always Finds a Reason to Go

This is where I think a fishing forecast can quietly become something different from what it was intended to be.

Imagine an engine that evaluates ten variables.

Seven are poor.

Three are favorable.

If the system is designed around getting me onto fish, there is a temptation to emphasize those three.

Maybe the run is weak, but the tide looks good.

Maybe catch rates have fallen, but tomorrow’s weather is excellent.

Maybe conditions have deteriorated, but another location still has opportunity.

There’s almost always something that can justify fishing.

Eventually the forecast stops being a decision tool.

It becomes a pressure amplifier.

It takes uncertainty — which naturally limits effort — and systematically removes reasons not to go.

That’s not what I want to build.

I don’t have to imagine it

In the middle of August I rescored a trip for a tester on a Tennessee reservoir. Four days, same water, same species. Every single day came back higher than it had a week earlier.

Nothing about the fishing had changed.

The first version had been scored five to eight days out, and the engine reduces its own confidence when it’s guessing that far ahead. When a real forecast arrived, that penalty came off. The result was a seven-point swing across four days with nothing behind it but the calendar.

Seven points is enough to move a day across a recommendation tier.

Now — that’s the engine being honest. Confidence should rise when the data gets better, and I’d defend the mechanism. But look at what it produced: four days that suddenly looked more worth fishing, when the water hadn’t changed at all.

He extended his trip.

I don’t think that was the wrong call, and the extra day he fished produced fish. But it’s the exact shape I’m describing. The engine found a reason to feel better about a trip, the angler acted on it, and the fish had no idea any of it had happened.

If a system that small can do that with one person, I’d like to understand it properly before it does it with a thousand.

Uncertainty Has Always Been Part of Fishing

For most of fishing history, uncertainty did some useful work.

You didn’t know exactly when the fish would arrive.

You didn’t know what happened twenty miles upstream yesterday.

You didn’t have hourly wind models, satellite imagery, real-time river gauges and thousands of anglers carrying cameras connected to the internet.

Sometimes you went anyway.

Sometimes you stayed home.

Sometimes you chose the wrong river.

Sometimes the fish got a break simply because nobody knew exactly where they were.

We aren’t going backward.

I don’t want to.

The information available to anglers today is extraordinary, and much of it makes fishing safer and more enjoyable.

But better information creates a responsibility for the tools interpreting it.

Not every uncertainty needs to be converted into opportunity.

Sometimes the Best Recommendation Is Somewhere Else

That changes how I think about WAIT.

WAIT doesn’t necessarily mean don’t fish.

It might mean don’t fish there.

And that distinction matters more than it sounds, because it is the only thing in this essay that could actually do something.

Everything above describes a problem. Redirection is a mechanism.

A forecast that says “conditions are poor, stay home” reduces effort. A forecast that says “conditions are poor here, and this other water is fishing well for a species you’d enjoy” moves effort. Those are different outcomes, and the second one is better for almost everyone involved.

The angler still goes fishing. That matters — I’m not trying to talk people out of a day on the water, and a tool that mostly tells you to stay home is a tool nobody opens twice.

But the pressure lands somewhere that can absorb it.

Maybe another species is a better choice — the lake is full of bass and everyone is fixated on the walleye.

Maybe a different piece of water has the same conditions and a quarter of the traffic.

Maybe tomorrow is meaningfully better than today, and waiting a day costs almost nothing.

Maybe the drive isn’t justified, and the water twenty minutes away is fine.

Concentration is the actual problem. Not that people fish — that everyone fishes the same place on the same day because the same information told them all the same thing.

A recommendation engine is unusually well placed to do something about that, because it is the thing producing the recommendation.

And occasionally it might simply mean:

Not this one.

There will be another tide.

Another run.

Another weekend.

Another fish.

The Obvious Objection

I should say the uncomfortable part out loud, because anyone reading this will get there on their own.

I’m building the thing I’m describing.

I’m writing a fishing forecast, publishing what it learns, and offering to score trips for strangers. Every argument in this essay applies to me at least as much as to anyone else — arguably more, because I’m doing it deliberately rather than by accident.

Two things make me think that’s defensible, and neither is a full answer.

The first is that this is happening regardless. The information exists. The models exist. People are already converging on the same water from the same reports, and they were doing it before I started. The question isn’t whether someone builds better fishing forecasts — it’s whether any of them are designed by somebody who has thought about this at all.

The second is that a tool which can produce WAIT can also produce somewhere else, and a tool that only maximises catch probability can’t. If concentration is the harm, then the only thing capable of redistributing effort is the thing producing the recommendations.

That may turn out to be a rationalisation. I’ve been wrong about plenty of smaller things while building this, and I’ve written most of them down.

But I’d rather be a person who built this while worrying about it than someone who built it without noticing.

What I Want the Engine to Optimize

Catch probability matters.

It has to.

A fishing tool that can’t help someone make better fishing decisions isn’t particularly useful.

But maximizing catch probability at every opportunity feels like the wrong objective.

I’d rather optimize for something closer to quality opportunity.

Good conditions.

Reasonable expectations.

Safe access.

A worthwhile experience.

And enough information to recognize when those things aren’t aligned.

That means the engine occasionally has to leave opportunity on the table.

It has to be comfortable producing WAIT when technically there are still fish available.

It should sometimes redirect effort instead of concentrating it.

And it should never manufacture a GO simply because that’s the answer I was hoping to receive.

Because if every improvement in fishing information ultimately produces the same recommendation —

Go now.

Go here.

The fish are here.

— we haven’t really built a better decision tool.

We’ve built a better way to apply pressure.

I want to build something smarter than that.

---

Want a sheet for your next trip? Any water in Washington State — tell me where and when.