Thursday, 17 May 2012

In search of randomocracy

It’s a tough decade to lead a major corporation or government. Things are changing so fast and the macroeconomic situation is so complex that it’s no surprise the finger gets pointed at Managers whenever things go wrong. The travails of new leaders at Nokia and now RIM and Yahoo have also placed scrutiny on the selection criteria for Managers.

It’s possibly because of this that there’s been an upsurge in research around a-periodical systems as a way of selecting leaders at all levels of organisations. Typically, this research is aimed at governmental systems and since “a-periodicity” is really a fancy way of saying “random”, the governmental system has come to be known as a “randomocracy”. Partly because I love the name and partly because I’m really interested in the practice of leadership in a digital, millennial business environment, I’ve been thinking about some randomocracy ideas could be applied to business management.

Random acts of kindness – the pure randomocracy

A pure randomocracy is a simple construct. Anyone in a country can be picked at random to serve in a national congress of consistent size, where they act with all the authority gifted to elected representatives today. Typically tenure is a period of a year, since this was the basis of the classical Greek system of government. God might not roll dice, but perhaps the Queen should?

The thesis that underpins the randomocracy is that provided the congress is sufficiently representative (a few hundred people, rather than a few tens) then a random sample of the population will be on average as successful in making effective decisions on national policy as a group that is selected through a structured process.

This is not actually as crazy as it sounds. Since the actual operation of a nation is carried out by civil servants, decision making is only a case of analysing given information and coming to a conclusion based on life experience and personal bias. The group then votes based on that individual bias or makes deals internally to support a portfolio of policies. Statistically speaking, over a period of years the will of the people will be done because the vox populus is actively represented at the point of national decisions.

Just to remind you, most legal systems in the democratic world work exactly this way. Decisions on the fate of individuals are made by a jury of their peers, randomly selected from the population. In some ways the UK’s House of Lords is also randomly selected, in that there is no elected representation and Peers often only have loose alignment to parties.

Hybrid random/ structured systems

A fascinating conclusion drawn by the authors of a recent paper on the subject was that a pure randomocracy isn’t worth pursuing as it is just as inefficient as the current, pure-elected system. That it isn’t any worse is telling! What the authors then went on to do is use the same mathematical model to work out the optimum degree of randomness for a government of a given size. Neat.

Their conclusion is that the most efficient system of government is one where a mix of elected, party-aligned people work together with a randomly selected group of independents. To my mind, this is a similar idea to that of coalition government, without the inherent weakness of that system, which is caused by alignment of formally structured small groups of people. The groupthink in those parties leads to amplification of bias and need to defend the herd, which in turn collapses the coalition in short order.
 Randomness in the workplace - the Dilbert school of Management

The authors of a Catania University paper on randomocracy have previously shown that randomly deciding promotions in a bureaucracy is at worst equally effective as implicit or explicit promotion systems. In many cases it will produce better results than formal processes because by promoting an underperforming person in an operational role, they free that role for someone more effective to step into it and improve overall productivity. Best not reflect too hard on what that says about the typical value-add of Managers.

On a similar train of thought, the practice of Strategic Management naturally contains an element of intuition and dumb luck. Rarely in my experience is one course of action so compelling that it’s the only option to pursue. Instead, there are typically several options that are equally good when judged on achievability and attractiveness. Chances are that rolling a dice to decide which to prioritise is at least as effective as trying to scientifically justify something that’s success depends on a complex set of externalities. Psychologically, however, I fear that it would be tough to align a business around a roll of a dice and execution is 90% of the battle.

In short, although probably workable, I suspect that these types of statistically valid, but culturally difficult systems are unworkable in most business environments. The possible exception to that rule are those organisations that are naturally replete with staff, such as governmental organisations, large outsourcers and similar. Statistics only work when there are sufficient sample sizes. Even so there is clearly strength in randomness and it shouldn’t be dismissed out of hand because it’s not the done thing.

Strength through randomness

Recently I’ve been working with a number of teams that when judged through the traditional lenses of depth strategic technique and client management skills were rather inexperienced. Yet over a period of 9 months, in a wide range of different client situations they’ve produced some of the most astoundingly brilliant strategy work I’ve seen in my career.

In thinking about randomness and reading the leaked Valve employee handbook cover-to-cover several times, I’ve come to some conclusions about why this is happening. Unsurprisingly, the first factor is having the right people. The second thing is to organise them in the right way. Not rocket science. No, this is much, much harder.

The teams I refer to above consist of a mix drawn from about 8 people, all of them displaying the typical top tier consulting characteristics of extremely high mental capacity, natural diligence, intellectual inquisitiveness and an open, sharing attitude. Within the team the mix is effectively random as utilisation is high and projects are not of uniform length – I’m doing roughly two at a time. A key thing, I think, is that at 31 I’m the elder statesman by 3 or 4 years. The youngest is about 23.

Why is the latter important? Because Generation Y people have yet to be corrupted by the sense of hierarchy that pervades traditional business. We have a lot of debate in these teams and it can get pretty heated. Everyone speaks their mind. Many companies claim that this is commonplace, but I’m really not so sure. In traditional structure (and I include most projects I’ve been on) there are leaders and there are doers. The difference here is that by working together for an extended period and by all being of exceptional calibre this is a group of friends as well as colleagues. Not many groups of friends have formal hierarchies – informal, yes, but that herd or tribal behaviour is a subtle undertone rather than explicit command and control.

As Valve describe it: “hierarchy is great for maintaining predictability and repeatability. It simplifies planning and makes it easier to control a large group of people from the top down, which is why military organizations rely on it so heavily. But when you’re (a) company that’s spent the last decade going out of its way to recruit the most intelligent, innovative, talented people on Earth, telling them to sit at a desk and do what they’re told obliterates 99 percent of their value. We want innovators, and that means maintaining an environment where they’ll flourish.”

What I’ve been witnessing is that flourishing of creativity and strategic innovation when talented people have the freedom to express themselves. No wonder our clients are happy right now...

Welcome to Flatland

Valve rightly make a lot of their flat structure as they represent an extreme of the above: it is an entirely flat company with no hierarchy, no official job roles. Whereas Google employees famously get their 5% time to do what they want, Valvers get 100% time to do what they think will make the company successful. And they are successful, having produced massive hit titles like Half Life, Left 4 Dead and the Steam online gaming platform.

Doing projects at Valve means convincing people to work with you. Employees can literally wheel their desk to a new team and start working with them whenever they choose. For this reason, the structure Valve have built is actually much more sophisticated than a “flatocracy” – it is a self-organising system. There’s no such thing as a Manager, just someone who the team nominates to co-ordinate the multiple legs of a project. Usefully, this type of system is also self sustaining as success breeds success, drawing in more talent and creating more value in a virtuous cycle.

That all sounds like paradise. Sadly it all depends on one thing: talent. Valve again: “Hiring well is the most important thing in the universe. Nothing else comes close. It’s more important than breathing.” Then they go onto say something very prescient: “when unchecked, people have a tendency to hire others who are lower-powered than themselves. We should hire people more capable than ourselves, not less. In some ways, hiring lower-powered people is a natural response to having so much work to get done. In these conditions, hiring someone who is at least capable seems to be smarter than not hiring anyone at all. But that’s actually a huge mistake.”

Anyone competing in complex markets would do well to remember this.

Appropriately, by searching for randomness, I’ve concluded something else!

Valve has been designed from the ground-up to be flat. Converting another business to act the same way would be very hard. Likewise randomocracy is too radical a departure for anyone to realistically consider (although Iceland has actually set up a commission of 950 randomly selected citizens to decide how to move on from the financial crisis – some would say that they had nothing to lose). But by taking extreme situations they suggest a couple of broad principles for succeeding in hyper-competitive digital markets:
  1. Decisions made by systems with process, structure and hierarchy (even if they are “best practice”) are in aggregate, statistically no better than rolling dice. Appreciating this fundamental truth enables you to relax the need for control and thereby create an environment where people have to adapt and innovate
  2. Recruit people that absolutely fit your culture and ethos, even if it means taking only one person when you think you need ten. Seek out people broader and smarter than you are
  3. If you truly believe you’ve done #2, then you should have nothing to fear letting any of your team lose on any problem. If you do have fear, then go back to #2!
This has been a very long post and possibly a little rambling. I hope, however, it’s useful. I think I’ve now come to understand something about how to treat my teams in the future. Let’s hope it works!

No comments:

Post a Comment