# The “Convergent” theory of finding truth in darkness

How do you know if your startup idea is a good one? Even after twenty customer interviews?

How do you know when to hang up the towel and try another idea?

The usual answers: It’s a balance. Trust your gut. But your gut is wrong so trust data. But you don’t have enough data so trust your gut. Don’t give up just because it’s hard. Don’t push if it’s too hard.

Follow this formula. The formula says there’s no one formula. The formula is to ask “the right” questions. The formula doesn’t know what to do with the answers. It’s different for everyone.

Until recently I haven’t had a good way to explain my idea of the answer. But recently I was rereading Feynman’s Lectures on Physics and, in one of those brief flashes of comprehension that comes when your mind is wandering yet active, I stumbled across the explanation.

The fun way to explain the answer is to paraphrase Feynman himself, because the point he makes is in itself fascinating and deep. And as you’ll see by example, it turns out to directly answer to the questions above.

It is an astounding fact that the force of drag on an airplane is proportional to the square of its velocity in air. That is:

`Fdrag = Cv2`

It’s amazing because it’s such a simple formula for such a fantastically complex environment. Not just the curving surfaces, but every bump on the wing splits and twists the air which then interacts with itself. The drag-inducing chaos of swirling fluid is so convoluted that even modern simulators can’t model it precisely.

It is almost unbelievable that an airplane in a wind tunnel reveals a characteristic constant `C` in such a simple law.

Humans love this sort of thing — emergent simplicity from multivariate chaos. There’s beauty in its brevity and power in its utility. We love it so much, the urge is in constant overdrive, and we see patterns and meaning even when there’s none.

We’re tempted, therefore, to call `Fdrag = Cv2` a “law” — a rule by which a mere human brain can get a handle on a phenomomon too complex for the fastest supercomputers.

But it turns out it’s not a law at all. It’s not power, it’s a tenuous coincidence, and not one of great utility.

How do we know this?

Because as soon as we try to understand similar situations using this law, we run into its limitations. If the airplane is flying slowly, it becomes completely inaccurate. If the airplane flies very fast it’s wrong again. If we make small perturbations on the wing the constant can change dramatically.

Perhaps most surprising is that if you physically remove one of the wings, that changes the drag on the remaining wing. There are forces inside and outside the aircraft, hidden to a casual observer, uncaptured by simple formulas.

Where is all this in our “law?” Nowhere. If we step outside our little cocoon, the law crumbles. It’s not a fundamental law, because it does not predict what happens in novel situations.

Contrast this to another so-called law — the Conservation of Energy — which states that the total energy of a system is constant. So if a ball falls in a gravitational field, as it loses potential energy it gains kinetic energy such that the total energy never changes.

Is this a true “law?” How can we tell?

We can make a complex series of ramps inside a vacuum, starting a ball at different heights and positions and letting it roll down and up and around, measuring the velocity the whole time. We find that the ball’s speed everywhere exactly ensures the two energies remain equal, regardless of the configuration of the ramp. This feels powerful — even in an arbitrary configuration, the law still accurately predicts the result.

With a real ball and a real ramp, friction slows the ball, thereby reducing total energy and therefore a violation of law! Ah, but we realize that “heat” is also energy — something we can measure and convert into other forms of energy — and when we measure the increase in heat in our experiment (in the ball and in the ramp) we find that the energy due to heat exactly replaces the energy lost as the ball slows, and again our law is correct. In fact, our law predicts how much heat, and we find exactly that amount, so now the law has just predicted the existence of new kinds of energy, and did it accurately, which is even more impressive!

Then from other experiments we have evidence that matter is in fact composed of gargantuan quantities of tiny objects, moving and colliding and vibrating. That suggests a different definition of heat itself — that it’s not a “new” form of energy at all, but rather the total kinetic energy from jiggling particles! Under this hypothesis we can make definite predictions about how much energy heat contains, how heat and particle density and pressure would have to change in a gas under various conditions, and so on, all on the sole basis that energy must be conserved, and in fact those predictions all prove accurate!

Even in the modern era with relativity bending and weaving time and space and quantum mechanics so strange that Feynman himself said that no one really understands it, still the conservation of energy has always been found to be perfectly correct.

With the drag-force equation, the deeper we dug the more we discovered that the “law” doesn’t encompass much truth; with the Conservation of Energy we found that the closer we look and the more we stretch ourselves the more powerful and simple the law becomes, the more applications we have for it, the more predictions it makes, and that is the characteristic of a bona fide fundamental law of the universe.

Truth in startups emerges or crumbles in the same way.

Specifically, before I validated the ideas behind WP Engine, I validated another idea for a startup. The key thing to notice is that during my customer development, everyone said “That’s a great idea, you should do it!”

Everyone.

Except, as I dug in with each person, the “truth” started diverging. One said I should target enterprises, charging \$1000/mo and selling through consultants. One said I should make it freemium and figure out how to make money converting 5% to \$5/mo. Another said charge a minimum of \$50/mo to cut out the moochers who email support but don’t pay for stuff. Another said the small-to-mid-sized business market is the untapped niche. One said I should use it to measure online ads and forget about measuring leads; the next said I should use it to feed leads to salesforce.com and forget about measuring marketing efforts; the next said I should use it to reveal marketing efficacy and not try to close leads.

Like the airplane law, I had discovered something intriguing, even exciting, but not something fundamental, not something with clear steps forward, not a Venn Diagram of ideas creating a large, dark area scintillating with lumps of variation, but rather a blotchy Venn Diagram with twenty lobes of dissonance.

But my experience vetting WP Engine was convergent. The more people I spoke with, the more agreement there was over the pain they had, whether my solution was an acceptable, and the the amount they were willing to spend. \$50/mo to make a site fast, scalable, secure, and someone who knows WordPress to answer the phone. Kick in a staging area and backups and it’s a done deal. Thirty of forty people agreed to sign up during their interview. (Twenty of the thirty later did.)

I had found the startup equivalent of a fundamental law — not an immutable physical law of course but something that behaves like truth — where multiple areas of inquisition lead to a common destination instead of leading to different planets. Whereas before I had an idea with the behavior of a curiosity with no power behind it.

Of course there’ll be no rubric to determine whether an idea is tenable or whether the situation is so bleak that you should give up. But in my experience this feeling of convergence or divergence is very strong if you’re being introspective and honest with yourself.

Your hardest battle is in fact with yourself, as you’re constantly tempted to bias the evidence in the most convenient direction (validation), and your fear of figuring out that your pet idea, while undeniably cool, is not a business, in the sense that other people don’t agree enough to give you money for it.

Just remember how expensive it is to blind yourself. Your project will meet the same end, only after a significantly larger investment in time, money, heart, and reputation.

### 17 responses to “The “Convergent” theory of finding truth in darkness”

1. Not that it matters much, but perhaps you want to change “conversation of energy” to “conservation of energy”.

2. Curt Dawkins says:

You can lead a horse to water, and even tell the horse how good the water is, but you can’t make him drink.  But good luck stopping a horse that wants to drink the water.   Sounds like your WP Engine customers are thirsty too!  Congrats on figuring that out.

3. Stephanie Hackney says:

I believe it’s important to note that humans are generally horrible at predicting their own future behavior.

So, even though many say they will buy, many won’t when the time comes to do so. Your own experience proves this out:
“Thirty of forty people agreed to sign up during their interview. (Twenty of the thirty later did.)”

That’s not to say that testing the “buyability” of an offering is not a good idea, and that people who claim they want what you’re offering won’t actually buy, just that testing is not fail proof. I have lived and learned this lesson…the hard way.

• That’s why I actually suggest you get the check from them, physically.  Then they have paid.  Sounds like it can’t be done but I know a dozen startups who did it.  Usually if it can’t be done it’s because in fact they don’t want it, and it’s a real signal.

4. I bet if you built that idea you rejected, it would be a success.

5. Mroberhozer says:

Valuable post (geekiness included), especially appreciate nailing every stereotypical response to the wall right from the get go.

Although, I’m left wondering why the “truth started diverging” on the initial idea. Were people blowing smoke by saying it was a great idea, then making an excuse by pointing elsewhere for customers? But I suppose in the end it doesn’t matter…if they won’t pay, who cares why.

• Good questions.

They weren’t blowing smoke about the root idea — it *was* good.  In fact, I find that most core ideas are “good” in the sense that someone has a pain and/or the technology is interesting — otherwise the founder wouldn’t also be stoked about it.

In that sense it’s like the first equation which *is* good in the sense that it’s useful in a certain context.  Nothing wrong with that.

But then people had different ideas about *why* it was good and therefore how to apply it, what features were still lacking before it was viable, which people really would pay and how much and why, and all that.

The kernel can be good but not have the trappings to be a business — all those other pieces.  In fact I believe most good ideas are like that.  Not enough other stuff.

There is an alternate ending, however, which would make sense.  That I picked ONE of those ways to make the idea truly a business, and did it.  That certainly could have worked!  It’s not that you have to throw away the idea.  Could have pivoted.

The reason I didn’t was simply that I wasn’t personally interested in those other ideas, nothing more.  Probably many of them could have been reasonably viable.

• Thanks, I think that’s a helpful distinction.
Took some fortitude to divorce yourself from that idea. I can only hope I can have the same objectivity and brutal self-honesty when the situation warrants.

6. wfjackson3 says:

Jason,

Holy crap, this is my top startup post of 2012 so far. Unbelievably good. Bravo.

• wfjackson3 says:

Jason,

Still so good six years later. Just asked a PM I work with to read it.

PS – Glad you are posting again. Hope you are doing well!

7. Your posts make me a little sad that I’m not smart enough to write like you do — but immensely glad that I’m, at least, smart enough to appreciate them!

8. Having a tough time with the Venn Diagram paragraph.  My sense is that it makes sense in the author’s imagination but I have encountered a challenge in rendering such a vision with my own.

Also.  It would be interesting to contextualize this analysis among other branches of science with other kinds of qualitative differences.  Here we consider the difference between thermodynamics and aeronautics (and fluid dynamics).  There are other sciences which focus on the mapping and analysis of complex ecosystems; these may be closer to our entrepreneurial science.

Great post.