“We didn’t do anything wrong, but somehow, we lost.”

— Former Nokia CEO Stephen Elop

Over the last few years, as the lean startup methodology gained traction amongst the Silicon Valley and startup types, an interesting idea that runs contrary to how things have been done in the past took root — because the lean startup methodology encourages founders to quickly iterate and seek feedback, the implied assumption is that keeping secrets are bad. The lean startup methodology’s argument against keeping secrets is one that makes rational sense: the last thing founders should do is burn tons on capital on a product they’re building in a dark when they don’t even know if people are going to use it or not.

Because of the rise of Twitter, Reddit and platforms like those on the Internet, it’s now much easier than ever to solicit feedback, the thinking goes, so why not take advantage of it and seek feedback for whatever you’re working on?

Likewise, in recent years, the Valley and the startup community at wide has rallied against the idea of making others sign a non-disclosure agreement before sharing their ideas. This is a good thing. But I think the Valley at wide has tipped the scales too much on the side of transparency. When the scales are tipped too much on the side of transparency, founders are less likely to try radical experiments — indeed, ones that often make a much more significant impact in the world if and when successful than an idea based on a cycle of rapid but marginal iteration.

Before I go into defending secrecy (or at least, reclaim some semblance of a balance between secrecy and transparency), there are a few bad reasons why founders shouldn’t opt for secrecy over transparency: fear that someone will steal your idea, fear that someone else will beat you to the punch, fear that someone will patent your idea, etc. If you opt for secrecy because of these reasons, you’re putting yourself — and your company — at a disadvantage.

But there’s one scenario where secrecy should triumph transparency: if and when it allows you to run experiments you otherwise would feel uncomfortable, or embarrassed to tell people about. With any experiment, there is — or should have — a 50% chance of failing. Such failures tend to hurt the company’s image and reputation, and give an opening to the press to mock them, investors an excuse to pass on an investment, clients a reason to choose a competitor, etc. Therefore, a company, being fearful of these pitfalls, choose to do what is safe, and not experiment — but unknowingly, initiate their decline into irrelevance and obscurity. This is where secrecy can give founders and companies the room to experiment, without having to be accountable to those who are not immediately associated with the project.

For many, the fear of reputational damage is enough to err on the side of secrecy. Look at Amazon: the Fire smartphone was a very public and humiliating failure, one that Jeff Bezos was interrogated for in the press, on earning calls and on stages in conferences (a little inquisition, if you will). However, the Echo was a smashing success — and Amazon did the entire project in secret, spanning 1,500 employees over years of labor. Look at Apple: there’s a reason why (beyond showmanship) Apple guards its secrets zealously. For every iPhone or iPad that gets released, one can safely assume that there were a thousand failures that never made it out of the lab. Not making those failures public means not having to be accountable for them to irrelevant observers — thus speeding the learning and moving on process up much faster.

There’s another compelling reason why startups should try to keep their experiments secret, even after they release it: it helps them skip past the early-tech adopters crowd and get direct, honest feedback from other people (normal; lives outside of Silicon Valley) who would make up the bulk of their customer base. Furthermore, not knowing the identity of someone behind a product you’re using (this is different from the D2C e-commerce business model where such established relationships are advantageous; this works best for software products) allows you to provide more honest and crucial feedback — given how quick many startups are given birth to and dies, such feedback can often mean the difference between life and death (“talk to your users” is, after all, one of Y Combinator’s motto).

Ultimately, the purpose of a startup is to do more, with much less resources than would be normally available to achieve such results. The easiest way to do so would be to experiment radically (thus allowing the learning to happen non-linearly)— allowing one successful experiment (out of many embarrassing and failed ones) to contribute an outsized impact to the goal and mission of the startup. Startups — and the founders that lead them — can’t fulfill this mandate if they feel compelled to announce what they’re trying all the time, and have to answer for them. Many of these experiments will fail, but that shouldn’t stop founders from experimenting. A company that stops doing radical things — out of fear of being held unnecessarily accountable  — is one that is already on the decline, even if they don’t know it yet. And that is why secrecy is okay.

TL;DR: Secrecy allows you to fail quickly and learn from those failures without having to go through a parade of shame and humiliation. That is necessary to produce outsized returns.