Let’s rewind back to 2013. I got my first ‘proper’ job at a plucky startup, and was introduced to the onboarding material, which was a mix of “watch these videos” and tribal knowledge. Some hidden gems that the modern internet has forgotten like WAT and MongoDB is Webscale (links at the end). There was also an apocryphal story about product development.
Once upon a time at a company far far away, an engineer was riding to work by bicycle. It was nearing the end of summer and there had been a cold snap, and when he got to work his hands were freezing.
Once he got to his computer he posted on the company chat about the idea of self-heating handgrips powered by a dynamo. A few others threw in ideas around adding a backup battery, and eventually the discussion escalated to a more general solution to instead shift body heat via liquid tubes with one way valves and a special jacket.
Eventually someone chipped in with a bit of wisdom, that they should all just wear gloves.
Throughout my time at that startup, just mentioning the word “gloves” was enough to head off overengineering at multiple turns. We achieved an amazing amount with very little resources, and this cultural aspect, in my mind, was a large contributor to that.
Fast forward a decade and a bit, 3 cities, 2 countries and a continent later, and I find I’ve now spent much of my career at deep-tech university spinoffs, mostly involving computer vision and robotics. The culture shock of seeing how much effort is put into the simple things I previously achieved in an afternoon has worn off, and it now seems normal to use Lie Groups and a fancy math library for code that I would have written by hand in a few minutes, albeit not taking the covariances into account correctly. The thing is, afterwards we tend to find that the covariances passed into the next component are mis-scaled anyways, so we have to hard code them. Or our Kalman Filter has some unobservable states and we tune the process noise so low that we might as well not be tracking that state at all. Truth be told, by the time we get to these conclusions we would have been better served using the simpler solution to begin with, and adding the complications once the problems actually hit.
I think a lot of it comes from the mindset of where the value lies. In a spinoff, the value, psychologically, is in the IP that is created, not the Product Market Fit. The PMF is something distant to be discovered, and left as an exercise to the present and future marketing and sales team, who are somehow expected to find the right problem for a solution that was already built. In contrast, if you are starting greenfield, trying to solve a problem directly for a customer (or perhaps a customer persona, hopefully realistic), the PMF is something the engineers themselves are inherently trying to solve as part of their solution, and the IP itself is less valuable as a standalone component, and less central to the culture. As an implication of this, there is a lack of understanding that stumbling across a strong business opportunity, even if it doesn’t align to the entire existing engineered system, is extremely valuable in its own right, and potentially even worthy of a pivot, regardless of whether your tech is working.
These aspects of overengineering and deferring the PMF until later, I like to refer to as Spinoff Disease.
Note the “over” in overengineering. This is a strong pejorative, and not just in the sense of “They should have spent their time elsewhere but at least we have this solution now.” I feel it is much more than that. Future development runs slower due to maintenance efforts. Reliability drops due to more moving parts. The core product offering is diluted away from where value is. This beautiful, useless engineering acts as a millstone around the neck of the team, constantly slowing down and distracting them from their core mission.
Another take I’ve heard on this topic, is getting a certain number of innovation tokens, and you have to choose where to spend them to have the most impact. If you don’t spend any, someone will come along and spend a few in the right spot and eat your lunch. If you try to spend too many, one of them will blow up spectacularly and you’ll waste all of your efforts in disaster management instead of building your product. The important part is spending them in the places where a huge success can give you a major advantage, and if they don’t pan out you can still find a workaround.
Of course this isn’t all cut and dry. Sometimes the company will get lucky, and a perfect opportunity opens up, allowing all of those overengineered capabilities to be used to their full extent. For example, geopolitics decided to give drones a second chance, this time as cheap munitions delivery platforms. This was so successful that there is a “Drones for Good” organization which tries to keep the non-lethal drone usecases in the public eye. And those manufacturers and suppliers who couldn’t get their reliability high enough for commercial usage, if their drones don’t make it back, well, that’s all part of the design now.
But the consumer dream that everyone who wants can have their own personal drone that watches them do parkour or documents their adventures, then automatically cuts and uploads it into viral videos? After drones enjoyed a brief honeymoon period, the “gloves” solution of 360-camera-on-a-stick is what we have seen motorcyclists, snowboarders, skydivers and deep-sea fishermen turn to, with the drone footage mostly relegated to being a beautiful albeit optional extra for some scenery shots with dramatic background music. That camera-on-a-stick option was obvious from the beginning in the drone communities, and given the skillsets they had anyways to push drones to where they were, would have been extremely quick to execute on. Unfortunately, most were tied to their vision and all that flight controller IP is a tough sunk cost to swallow when thinking about “gloves”.
As promised:
Oh, and it turns out there have been a few heated grips for bicycles which have now showed up on the market in the last decade. For me, I’m sticking with gloves.