Why we're building Ethyca

If a structural engineer wore a t-shirt that said "move fast and break things," you would never set foot on a bridge they built. In every real engineering discipline, the person who designs the thing is responsible for what it does when people depend on it, and that responsibility is not negotiable, not something to be deferred, and not something you get to opt out of because you were moving quickly. We accept this everywhere, in bridges, buildings, aircraft, and medicine. And then we get to software, which now runs more of the world than any of those, and we decided the rules don't apply.

We call ourselves engineers because we wanted the title. But no other engineering discipline tolerates what software has normalized, shipping things that fail, at scale, into the lives of millions, and treating the failure as a cost of speed rather than a failure of responsibility. Imagine our bridge collapses and we blame the lawyers, argue about which contract covers it, which regulation was breached. If a real bridge collapsed, no one would ask what the contract said, they would ask who designed it. That instinct, that the people who build the thing are answerable for it, and answerable against a standard, not just their own judgment, has gone missing exactly where it now matters most. The infrastructure society actually runs on is no longer steel and concrete, it's code, and it's increasingly code that makes its own decisions, at machine speed, with no person in the path.

I have two young daughters. The infrastructure being built right now, by engineers who've been told that breaking things is just what moving fast looks like, is the infrastructure they will live in and on their whole lives. I am not willing to leave whether it can be trusted to chance, and I don't think being a parent makes me sentimental about this. I think it makes me clear-eyed. Someone is going to build the systems our children depend on, and the only question is whether the people building them believe they're responsible for what those systems do.

I want to be clear about what I'm claiming, because it's easy to hear this as a complaint about technology, and it isn't. I am a technology optimist, I believe, genuinely, that software is one of the great forces for good we have ever built, and that almost everything worth doing in the next fifty years will be done faster and better because of it. The state of our public conversation, our collective attention, the erosion of trust, none of these were acts of nature; they were downstream of specific engineering decisions, made by talented people who had decided the consequences were someone else's problem. That is the thing I refuse to accept. Not that technology is dangerous, but that the people best placed to build it responsibly have been taught that responsibility isn't theirs to hold.

That is the conviction Ethyca was built on.

I know how the word "governance" sounds and I've often joked about it, to defuse that tension. It sounds like the opposite of everything a good engineer wants to do, like committees and sign-offs and the thing that shows up to slow you down after the interesting work is finished. For most of its history that's exactly what it has been, and if that were what we were building I wouldn't have given seven years to it. But the reason software got to be reckless in the first place is that being careful was slow, and being slow was a real cost, so we told ourselves speed and responsibility were a trade and picked speed. Runtime governance is the argument that they were never actually a trade. The harness is not what slows the machine down; it's what lets you run it flat out in the right direction. Done right, this is the thing that makes moving fast safe, which is the only way moving fast was ever defensible.

Everything an organization decides about how it will and won't behave, its commitments, its principles, the lines it will not cross, has always lived in documents and in people's heads, and been applied by a person who happened to be in the room. That worked when the world moved at the speed of people, it does not work now. Machines use data on their own, make decisions on their own, at a speed and scale no person can stand in front of. So the gap between what a company decided and what its systems actually do has become the defining risk of this era, and almost nobody is working on it. The entire industry is racing to make machines more capable. Very few people are asking whether the institutions deploying them can still do what they said they would.

This vital question, whether an institution can hold to its own word at machine speed, is the one we've chosen to work on. It isn't only a compliance problem, and it was never only about privacy, even though privacy is where we started and where it still matters most. It's the problem of giving software engineering the thing every other engineering discipline already has, a way to build fast and still be answerable for what you built. A harness that focuses but never slows. The ability to take everything a company has decided is right, and make it actually govern what its machines do, everywhere, every time, at the speed they now move.

Everything we’ve built at Ethyca over the last seven years has been in pursuit of this idea. It has taken time to understand the problem well enough to explain it simply, but this is the clearest account I can give of what we’re building and why. Until now, much of that thinking has been implicit in the products we’ve built rather than written down. This is an attempt to change that.

I don't believe doing this well and building a serious, valuable company are in tension, and I have no patience for the idea that they are. This isn't a charity or cause with a business attached. It's a real business, solving a real problem, that happens to matter enormously, and I think the companies that will matter most over the next decade are exactly the ones that refuse to choose between those things. We intend to be one of them.

We have, I believe, about a decade before the harness for machine-speed systems is either built or becomes dramatically harder to build, before the way machines make decisions on our behalf hardens into infrastructure we can no longer easily change. That window is open now but it will not stay open. What gets decided now, whether the systems our institutions run on can be held to the commitments we made, is going to shape the world my daughters grow up in and the one yours will too. That is the thing we are building – it is worth building well, and it is worth building fast, and the thing I most want you to take from this is that those were never opposed.

The last generation of software was built by people who decided they had to pick one. We don't, and we won't.

-- Cillian Kieran