A verdict, not a dashboard: one answer you can trust

Open any modern marine weather app. Windy, Surfline, Fishbrain, Navionics, pick your favorite. What you see first is a wall of numbers. Wind speed and direction. Swell height. Swell period. Wind wave. Water temperature. Barometric pressure. Tide chart. Moon phase. Sunrise. Sunset. Solunar majors. Solunar minors. Current speed. Current direction. SST anomaly. Chlorophyll.

The user’s job is to look at all of those and come to a decision: is today worth going?

This is the design failure at the heart of every ocean conditions app.

The actual question everyone is asking

Go read any “what app do you use?” thread on Spearboard, DeeperBlue, The Hull Truth, or Hull Truth’s offshore section. The words people reach for are remarkably consistent:

“Is it worth the drive?”
“Is the bite on?”
“Are conditions firing?”
“Did we get a window?”

These are not wind direction questions. They are decision questions. The user has already decided they want to get on the water. They’re deciding whether today is the day.

What they actually want is one answer they can trust about today. What they get is 17 numbers.

Why every app ships a dashboard

Dashboards exist for good reasons. The data really is uncertain. Different users care about different signals. The app doesn’t know your skill level or your boat or your willingness to fight a short choppy window. Handing you all the data feels respectful of your judgment.

But this respect is also a dodge. A dashboard means the app never has to commit. If conditions are bad and you went anyway, the dashboard showed you the wind was 20 kts, so it’s on you for going. If conditions were great and you stayed home, the dashboard showed you the flat swell, so you should have looked harder.

A dashboard is an app that refuses to have an opinion. An opinion is what you actually want.

What “just tell me” actually requires

It’s one thing to complain about dashboards. It’s another to build something opinionated enough to say not today, and here’s why without being wrong so often that users stop trusting it.

Three things have to be true for an opinionated verdict to work:

First, it has to read the same signals the expert is reading. A good surf forecaster weighs swell period above swell height in their head. An experienced spearo weighs the rain from the last three days almost as heavily as the current wind. An inshore angler checks the tide and the light before they even look at wind. The app has to internalize those weightings (per activity) or its verdict will be laughable.

Second, it has to express its uncertainty honestly. There’s a real difference between a bare “GOOD” and “GOOD, but the wave height row is the weak one and the visibility estimate is only at 40% confidence.” An opinionated app that hides its uncertainty is worse than a dashboard. A good verdict shows you where it’s unsure.

Third, it has to be wrong out loud. When a verdict turns out wrong (and some of them will) the app has to say so, explain why, and close the loop. Users will forgive wrong but transparent. They will not forgive confidently wrong.

The tradeoff we accepted

The downside of an opinionated app is that one wrong call is worse than 17 right numbers. A user who reads “GOOD” on a blown out day and drives three hours loses faith fast. If the verdict is wrong even 10% of the time and 10% of those wrongs are on the days people drive three hours, that’s a permanent trust break.

So we accept a design tax: we show where the verdict is unsure instead of hiding it. Every factor shows its score and the reason behind it, the visibility estimate carries its confidence, and the dialog lists where the data comes from. Where we can’t score part of the picture, like the shore bearing a surf verdict would need, the app marks that verdict partial. The alternative, confidently wrong to save face, is how trust dies.

The signals that actually matter, by activity

One thing we learned building this: the weightings really do change by activity, not in a small way but drastically.

Spearfishing and diving care enormously about water clarity. Nothing else matters if you can’t see: no amount of flat wind and moving tide will compensate for 2 feet of vis. This is the top signal for both, and every other signal is secondary.
Rod and reel fishing leans hardest on the tide: moving water, a decent range and a turn coming up all score higher, and spring tides around the new and full moon add a little more. Wind sets the envelope (too windy and you can’t fish it properly). Feeding is a separate read, the bite score, and it weighs light first (the hours around sunrise and sunset), then how fast the tide is running, with moonlight a small nudge at night and cold or hot water extremes pulling it down.
Boating weighs wind and waves about equally; sailing leans on wind first. Tide matters a little for boating, around inlets. Moon matters not at all.
Surfing weighs wave height, swell period and wind speed. It has no shore bearing, so the app marks its verdict partial.
Freediving puts visibility first, then waves and wind.
Kayak/SUP leans on wave height first, with a hard ceiling, then wind.

A single verdict that ignores these differences is either useless for everyone or right only by accident.

Why “AI powered” is a tell

When a conditions app pitches itself as AI powered, that’s usually a confession that the app couldn’t articulate its factors. The category learned this the hard way: Fishbrain’s BiteTime has been a source of debate for years, and forum consensus is that users can’t tell if it’s better than a coin flip, partly because the model is opaque. A GPS collar study found solunar theory was “accurate about 25 percent of the time within a three-hour window.” And yet solunar tables still ship at the core of most bite prediction products, because it feels rigorous.

Opaque “AI” verdicts don’t solve the dashboard problem. They just replace 17 numbers with one number and no reasoning. The user still has no information about whether to trust the output. That’s worse than a dashboard: at least a dashboard lets you audit.

The alternative is transparent opinionated. The verdict is a single number. Tap it and you see each factor’s score out of 10 and the reason behind it. In Fishing mode that might read: tide is an 8.5 because it’s running and turns within the next 90 minutes, moon is an 8.5 because it’s a spring tide, wind is a 5 because it’s blowing 15 kts. Now the user has both the verdict and the audit. They can override us when they know something we don’t. They can learn from the factor weights what signals actually drive conditions in their area.

What this changes for users

If you’ve been stitching together three or four apps (Windy for the synoptic wind, a tide app for the tide, Fishbrain for solunar) stop. That stack exists because you’re combining apps built for different jobs, and the weighting ends up in your head, after looking at 30 numbers across three UIs. That’s tiring, and it’s easy to get wrong.

You’ve been doing it for years. That’s what Submarius is trying to change. One app. One verdict. Every factor tappable. All of the weighting is done for you, per activity. No stitching. And when we don’t know, we say so.

Where we might still be wrong

There are cases where a dashboard will always win: professionals who actually want to look at the raw models (offshore tournament captains comparing ECMWF vs. GFS to pick a route, commercial fishermen with decades of local knowledge). Those users aren’t going to stop running Windy or PredictWind, and they shouldn’t. The dashboard is the right answer for them.

But for everyone else (the weekend spearo, the club diver, the inshore angler, the coastal boater) a good opinionated verdict is a structural improvement. The alternative we’ve been living with isn’t freedom of choice. It’s homework.

The question isn’t whether dashboards should exist. They should. The question is why every ocean app ships one by default when most users are asking a decision question.

Probably because it’s hard to do the opinionated version well. We’re giving it a shot.