10 Ways That Organisations Die
One of my previous posts talked about a few ways in which organisations often end up falling into an inefficiency trap. There was a considerable amount of discussion about why this could happen. Since one of the obsessions I've explored on this blog is re the constituents of organisational design and how some choices leave you in unenviable failure mode, this is what I learnt from creating a 2x2 to explain why large organizations suck at innovation.
Robin Hanson had a story about how this derailed even the small group of researchers working together. The paths to organisational failure are many, and it's easy to get lost in specifics here, but we should at least attempt a map with a “here be dragons” rather than throw up our hands and get lost in instrumental variables!
It took Western taxonomists more than 80 years to classify the Platypus. And since companies are roughly as complicated, here's something to get organisational taxonomists started on theories as to why they suffer from the inexorable power of inertial decay.
Mission dilution: As existing employees who are highly bought into the company's vision and/or mission get replaced over time, the level of "norm following" gets depleted. Everything moves towards increased systematisation to make up for the lack of more informal connections, which only increases bureaucracy over time since not everything can be codified, and creates a destructive spiral.
Governance problems: Perhaps it's the fact that most organisations don't actually follow a strict command and control protocol, with clear leadership and easy to follow instructions. The muddling around brought about by independent thinking middle managers creates more chaos than desirable, and puts up roadblocks where none should exist. When Steve Jobs wanted rounded bezels, mountains were moved to make it happen. When Robert Moses wanted to pave neighbourhoods to put in freeways, it just happened. That requires some firmness that doesn't exist in most organisations today.
Talent dilution: Larger organisations have a much lower bar in how they attract and retain talent. It's not lower as a rule, it's lower because of volume. Once you need to hire a 1000 people a hire, your policies start becoming a tad more expansive. As a result they are just much less capable than the upstart startups who are hungrier, more eager, and just smarter at solving problems. The lack of talent is the real issue.
Ambition (Thinking too small): Everyone gets much more risk averse once there is actually something to lose. This percolates through the system till there is a large enough array of ways in which ideas can get shot down. What used to take 1 yes now takes 6, and 6 yeses are exponentially harder to get. Creating failsafe plans becomes key focus of all executives, and the whole organisation becomes much more risk averse.
Training and proceduralisation: As organisations grow, the burden of each older generation to teach the newer generation grows. This growth can only be solved through codification of existing (often amorphous) knowledge. Codification is hard though, because most jobs aren't easily put down on paper. And of course once codified, you have to follow the rules set out for you. Which creates rigidity, bureaucracy and all the failures of innovation. And when someone smart enough tries to break out, Chesterton's Fence comes back to bite you too, putting the final ignominy in a tiresome game.
Process inertia: Larger organisations have a tremendous amount of things which they are in charge of maintaining. This means that the lion's share of their efforts will be spent in maintaining their existing operations rather than creating new growth options. When two thirds of your budget focuses on keeping the lights on, you end up not worrying so much about innovation.
Analysis vs action: Larger organisations have more people since they have more things to do. By definition this also means more managers and more layers of bosses. This creates massive analytical needs for the org, as opposed to smaller orgs which are more optimised for action. This increased analytical burden ("what should we do?"). And what would've happened instantly in a startup ends up becoming a six people decision making chain with each person creating and summarising a report for the person above. This rather obviously acts as a brake on any new initiative.
Information blockage: In larger organisations there is simply too many people across too many levels who all need to be brought onto the same page before deciding things. And our information transmission rates are too low to do this properly, so there needs to be, by necessity, a high degree of summarisation. This summarisation creates errors and distortions, and ultimately hobbles the organisation as a whole.
Distorted incentives: The problem is that within larger organisations the incentives for most of the employees is highly twisted. Unlike the payoff when you start your own company, here you don't get to get rich, or even recognised. You might have the best ideas but while "intrapreneur" is a word, it just means all the work of a startup, all the annoyance of a large organisation, and none of the benefits. There's no real payoff in trying to be too efficient or innovative.
Broken feedback loops and boondoggles: Within a large organisation, if you want to get something started and built, it'll take a decade if you're lucky. Oddly enough, for most companies, you don't get the kind of resources you can get if you're a startup, and you don't get recognition along the way. So while a startup can raise $300m and build a multi-billion dollar company (while still having a meaningful chance of failure) which would've been great to have, you simply can't pour that kind of resources into something which has a 80% chance of failure. Funnily enough while we do see giant boondoggles (again, hello Google), neither their funding profiles nor their untimely terminations seem much in line with individual or team will, instead seeming like the capricious actions of an uncaring god.
A few meta reasons to explain organisational failure modes.
A couple other ideas that emerged from the feedback from this article was on whether we're just looking at the problem all wrong. Any problems suggested on the lack on innovation on the part of the larger companies is a selection bias issue. After all, we do have 1000s of smaller companies that don't do anything innovative and go bust too. So indexing on the small number of startups that seem to actually produce something of value is just cherry picking.
This concern is valid. The counterpoint being that the organisational ills I wrote about wasn't just to suggest large companies suck (or seemingly suck). It was the fact that they are often institutionally incapable of innovation. They don't even take advantage of the actual R&D innovation built inside their own company! Most of the tech industry is like this. It's practically the founding myth of Silicon Valley that much like Prometheus, sans a non-compete or two, some employees skip an old behemoth and take their new whippersnapper ideas to start a competing company.
And then you look at companies that were considered highly innovative, like Google, and see how within a few short years they've managed to climb back down that innovation ladder to become yet another ad merchant. My question is, isn't that just weird? Shouldn't we see some stickiness in terms of the time series of innovation? The half-life just seems way too short!
A thought was that this is simply reversion to the mean. Every now and then there's outperformance, but the outperformance quickly reverts back, and that doesn't mean anything in particular is going wrong, just that it was an unsustainable trend to begin with.
However I've started to become more skeptical of reversion to the mean arguments as they're increasingly over-applied. It's most interesting when the underlying distribution is essentially random. In this instance, this concern makes it sound like success in innovation is essentially random. And is that true? I'm not sure. There are companies that have managed to move with the times and live for more than a century, like GE. They are not trillion dollar powerhouses, but not the equivalents of a thousand year old ryokans in Japan either. But their very mediocrity makes them not seem particularly innovative in any case.
Also, if innovation is essentially just random, ideas that either spark or don't, then the point in having larger R&D departments is purely to get more shots on the goal. If that was the case, corporate R&D should resemble a good VC portfolio, with a bunch of crazy ideas and moonshots, but that doesn't seem to be the case for the most part!
I still remain confused about this topic. I wish there really was a science of studying organisational success and failure with some rigour. Perhaps I’m just missing the literature, and if any of you have ideas I'd love to hear them!
There has been some research on what makes organisations innovative from post WW2 defence departments like this book: https://www.amazon.com/Sidewinder-Creative-Missile-Development-China/dp/1591149819
They studied the success of OSRD during WW2 and came up with some principles. To put them simply they were:
1. Innovation occurs only in small group of people (15-30), any larger number and the innovative ability of the group decreases.
2. Innovation requires a diverse group of individuals. They don't mean diverse in the modern blacks, latino, female sense. They meant diverse as in a missile development team should have a shop floor mechanic, theoretical physicist, aerodynamics expert etc, as many viewpoints relevant to the problem as possible.
3. Innovation requires a flat hierarchy, anyone should feel free to share their ideas to the boss and the boss should be capable of finding which of them are worth pursuing
4. Innovative organisations require rapid prototyping and external contracting. Anything unrelated to the problem they want to solve should be sent to outside contractors to solve so that they don't waste time in it, and rapid continuous prototyping is better than time spent excessive planning on the drawing board.
5. Innovative organisations require a large risk tolerance. They shouldn't forever be in need of money, ideally they should have enough money to fail a couple of times without worrying about closing down.
Apart from this I would say expecting Google as the large organisation being innovative is unrealistic, but small groups within google are quite innovative. Google is no bell labs but Google has invented Tensorflow, Waymo, Alpha-Zero and maybe something in Quantum Computing in recent history so it's no slouch either. I also suspect having a good 'taste' in choosing problems is important. Regardless of how amazing your group, if you end up choosing an incremental problem or too outlandish a problem, both will end up as a waste of time.