IT#10 Office Evolution and Unnatural Selection

#KnowledgeEconomy #DiscoveryEconomy #ITManagement #Management #SocialNetworkEffects #UniversalEvolution

This article explains why any knowledge economy company represents an evolutionary environment, why this is important, and how it works with the factors which affect the market fit of your company. The topic is most likely of interest for managers or CEOs. This is done using a computer simulation model.

Really? Office Evolution?

Yes. Check other articles on this site. There is a theory of universal evolution which applies generic principles of evolution to non-biological systems. It says that you have evolutionary environment whenever you have:

1. A limited pool of resources, e.g. an ecological niche, audience or salary fund.

2. A number of entities that compete for this pool of resources.

3. Eventual demise and reproduction of these entities with sort of inheritance and mutations.

A typical business satisfies all these conditions. It has:

1. A limited salary and bonus fund.

2. Employees compete for it.

3. Some employees are getting fired, some leave on their own for better pastures, and hiring is done through an interview system, which provides some similarity of those hired to those who interviewed them, with required mutations on the way.

So, if you are a manager, you’ve got an evolutionary environment. Congratulations. For more details see the article on this blog “Corporate parasites”, there is enough on this topic there.

Why do I care?

Maybe you don’t. If you are a member of this evolutionary environment, one of those competing for funds and raises, ok, it’s hardly new for you. If you are in a managing position, it also depends. Slave owners of anticity hardly cared about the skirmishes between their slaves. A noble duke, marquis or count hardly cared about the competition between their peasants. Even for a factory owner relationships between the workers were hardly of any concern.

So, what has changed? Why is it important now?

The technology. The people needed to handle that technology. The ways to manage the people needed to that technology. Or to put it short, the Knowledge Age and the Knowledge Economy.

With the Knowledge Age the knowledge to control the production moves from management to the employees, That’s the very definition of the knowledge economy and knowledge workers. The knowledge how to produce the product, component or service just does not fit into a single brain anymore. So, if in the previous ages it was the responsibility of a slave driver or factory floor supervisor, in the knowledge economy it’s the responsibility of the actual workers. Your managers won’t even understand that something goes wrong until too late. So, now it becomes very important that you have the right people.

As mentioned before, this is only applicable to the knowledge economy businesses, like IT shops, Internet services, software development, and some other creative professions. And if you are in one of those, you need to read this article.

The model

Below you will see a computer simulation model that computes possible market fit of the company based on the parameter that you as a manager may try to control. The model is simplified for ease perception. Each person is represented with a colored dot. The color represents the person’s market position. At the same time, we have an ideal market match shown on the right.

Timeline is split into generations. A generation is an abstract period of time that could be a quarter or a year, this does not matter. What’s important, it reflects the changes in the staff with the people who left replaced with the people hired. Each generation takes a row of dots, and they move from top down.

Since the model has a lot of parameters, I’ve picked a few preselected to demonstrate the most important conclusions that you may see from this simulation.

That’s it. I will introduce the simulation model parameters with screenshots showing their impact on your team’s culture.

Let’s start

The first and most important parameter is attrition. It includes the people whom you have fired, as well as the people who left your company on their own. And the first value of attrition to try is zero. Ok, you don’t fire anyone, no one leaves the company. Sort of a corporate employee heaven. What will be the market fit of your company?

Well, you just stay with the same random people, so whatever market fit you had at the beginning, you have all the time. If you are lucky, it could be ok. If not, tough luck.

Now let’s consider another extreme, sort of a corporate hell. Each generation everyone is fired, and the replacement is hired at random. As you would guess, this does not help much.

Both examples above don’t exist in the real life, but let's check the metrics:

• Heaven model:

  • Hiring costs: 0

  • Market fit, random, 26% misfit in the case on the picture.

• Hell model:

  • Hiring costs: almost 250K (we measure in abstract units, so we have a mushroom instead of a currency symbol)

  • Market fit, random, 31% misfit in the case on the picture.

Now, let’s try something moderate and not too harsh, like 0.3% attrition.

As you see, hiring costs are very moderate, around 500 🍄, and even then the market fit is much better, in case of the shown run misfit is around 15%. However, if you look at the picture, it's still not so good.

Maybe we can get attrition much higher, e.g. 50%, and get a better fit?

Nope. The hiring costs jumped to a huge 125K, and market fit is even worse, around 20% misfit. Which is kind of explainable, since at this rate you begin to fire perfectly fit people and replace them with a noisy hiring process.

General management theory says that a good attrition rate is around 10%, and the picture below confirms that:

The target fit becomes very good, around 2%, while hiring costs may be not ideal, but fairly reasonable, around 25K.

Notice that so far, we only fired for mismatch but hired pretty much randomly. This makes sense since very few companies were able to set up a really working hiring process, and even then, it mostly filters out totally incompetent people, and hiring errors are common. That’s true for large companies like Google as well as small and medium companies. People are just not so good at it. That makes managers oversensitive to the threat of hiring a wrong person, but that does not help either.

However, let’s imagine that:

• we know the market, and

• we trained some AI to see if a person is a fit to the market.

Of course, for now the second assumption is pure Shi-Fi and the first one goes well beyond it. But nevertheless.

O-oh! What a beautiful picture would we have in that ideal world! Alas, we don’t live in the ideal world. And in the real world, at least research shows so, people hire those whom they like, not those who are best for the job. This is actually the main reason why perfect hiring is so hard. Anyway, here is what we get:

You see, the market misfit has grown again to around 20%, and the color is totally grey, no matter what the market target is. Why? Because instead of market we follow the average opinion in the company, the opinion of majority, and for colors the average majority opinion is always gray. That's easy, we started with random colors. Each channel, red, green, and blue, has values from 0 to 255. What’s the average? 127.5, or hex 0x7F. What is the color for 007F7F7F? Right, grey. The majority opinion is always grey.

Now let’s return to the balanced model but add managements errors. More precisely, errors when firing at 20% (if you will play with the code, you can set it for hiring as well):

Surprisingly, it’s not that bad. The market misfit is around 9%, much worse than a balanced model, but still not bad and better than many other considered scenarios.

Now it’s time to come to a very interesting parameter: mutations. The difference between management errors and mutations is that with management error it does not make a difference whether the person is a good fit or not. Basically, you may fire your best or hire a worst. With mutations you hire someone close to your goal whether it’s a market target like in the picture below or a majority opinion. Here it is:

As you see, the market fit is still not that bad. Clearly mutations in hiring don’t affect the outcome much. Of course, it should not be completely random, but some mutations are totally ok.

Now we get to one more model parameter. The economy is not fixed, it changes, and often it changes suddenly. For that we have “Change target every N generations”.

As you see, if we hire by target, everything is beautiful. The company adapts really fast. Even addition of mutations, does not break that picture much:

Notice, that is you try hiring by the majority opinion, you will see the same picture above that you had without the market changes. The company run by the majority simply ignore the market.

Now, many companies in economic downturns decide to lay people off. I have a separate topic about layoffs in the Knowledge Economy, but here we have a much narrower look: do layoffs help with the market fit? The answer is “no”.

Even mutations don't change the picture:

Again, since layoffs are mostly random, and in some countries that’s by law, they may change by accident the majority opinion in your company, but will it help the market fit? Unlikely.

Now we get to the last and actually most interesting parameter of the model: market pressure. What does it mean? It means that the more distance from your majority opinion to the actual market target, the more the market punishes you. Normally what happens is that your sales drop down. But since the model does not include sales, we will skip this step and start reducing the number of employees.

Not bad and usually recoverable if you fire or hire by the market fit. But try to do that by majority opinion and:

I ran several hundred simulations for this scenario in the hope of seeing exceptions, but no. The result was the same. If your company moved by the Aristotle curve to the point where it cannot listen to the market, and there is (as always in real life) market pressure, your company is doomed.

Conclusion

First of all, some caveat emptor warning: market fit is not everything to run a good business. You need some diversity, but that’s a subject for an entirely different article. Anyway, don’t worry. Even if you try to get rid of some diversity, then unless you are an early Apple under Steve Jobs, you cannot avoid it. So, no problem.

Still, it’s very good to be not too far off the point. And for that we see that:

1. High attrition rate and low attrition rate are equally bad. Something around 10% seems the best, but since the period of time covered by one generation is not defined that number is not very useful. This number was taken from some management theory papers, and there it was an attrition per year.

2. At least hiring of firing has to have some resemblance of market targeting.

3. Involuntary part of attrition (firing people) seems to affect the market fit the same way whether it’s based on the market target, majority opinion or completely random. However, if your hiring process is really bad, firing by market misfit may help, albeit not quite achievable (again, matter for another article).

4. Hiring by market target make your company fit the market very fast. The problem is how to do that.

5. Hiring by majority opinion (what normally happens) cannot improve your company market fit. You need some other methods to improve it.

6. Sharp changes in the economy are ignored if you hire by majority opinion.

7. Layoffs don’t help market fit.

8. If you have to reduce the company size due to the economic situation, try to lay off by market fit, e.g. the projects, that your company does not need. Caveat: while it’s easier to see the projects market fit than one of an individual, this still may be not that easy.

9. If economic downturns force you to reduce the company size, doing things bu majority opinion may kill your business.

I have some ideas how to enrich this model more, e.g. assigning different weights to different people’s opinion like it happens in real life, but that’s the matter of a different article.