notes-institutionalDesign-antiPatterns

profitsharing fiefdom

Back in 'the old days', an institution is organized to maximize autonomy while also maximizing quality. This was done by structuring it as a federation of independent individuals ("masters"), who would each be responsible for finding things to work on and for doing the work. The fruits of the labor of each master would mostly go to that master, although a little bit would be 'taxed' to support central organizational overhead. The central organization would provide administrative and a few other centralized services, as well as prestige (because the centralized brand is the sum of the brands of the members). Of course, there were also apprentices, who worked under the masters, working on what the masters told them to. They were salaried and would not share in any further fruits of the labor, but that was okay as they would eventually become masters themselves; they were 'paying their dues'.

The problem comes when, as the institution grows, they allow the ratio of apprentices to masters to grow and grow. A critical threshold is reached when a competitive intermediate rank is created. The intermediate rank is above apprentice but below master; but crucially, they are given neither the autonomy nor the profitsharing of the master. The intermediate rank is described as somewhat independent, but is ultimately dependent upon the master for their job.

Sometimes the intermediate rank is justified as "First we have to train the apprentices, and only after they have been trained can we begin to evaluate them to see what they do with it". This would be fine if it were the whole story, but often the real reason is a reluctance to expand the number of masters in proportion to the amount of work to be done. Here's how to tell: Are there many more apprentices and intermediates than masters? Is it rare to see masters working on projects themselves without a team of apprentices and intermediates on the same project? Would a master who refused to take on apprentices and intermediates be considered unproductive, as opposed to merely shirking a duty to train the next generation? Do masters frequently delegate quality control to intermediates? Do masters who have large teams of productive apprentices and intermediates accumulate institutional power due to the sum productivity of their subordinates? Is there a real danger that some masters will hold back good apprentices and intermediates because they are concerned about the loss of productivity to their team when they move on? Is the average time spent as apprentice and intermediate longer than would be realistically needed to decide if they are qualified for promotion? Is there a sense that the principal barrier to promotion to master is not the readiness of the candidate but rather waiting for a spot to open up? Does the population of intermediates grow with the amount of work to be done, whereas the population of masters is limited by other factors?

The system was always exploitative in that the master has an incentive to overwork the apprentice (because the master benefits from the rewards of accomplishment within their group but does not suffer the effects of overwork). However, before the creation of the intermediate rank, this was seen as a necessary evil because no one wanted to trust a mere apprentice. Note: this is just a fairy tale history; in real life there were often intermediate ranks such as "journeyman" even in the 'old days'.

In addition to this moral failing, the new system has become less attractive to apprentices. Before, those who sought autonomy would be attracted to this career path; after paying one's dues, one becomes autonomous. The situation of the apprentice may have been unpleasant, but the end was in sight. Now, those who seek autonomy are repelled by it; years of labor, and the best one can achieve is to be promoted to intermediate, upon which the whole unpleasant cycle merely repeats.

A metric for how far the antipattern has progressed is to look at the average age at which a person attains autonomy (that is, the rank of master). The result to be avoided is a system in which, by the time one achieves autonomy, one is old. The goal should be for individuals to achieve autonomy before they have reached 'middle age', whatever that means.

In addition to these failings, the system can be inefficient. In cases where this pattern arises in a guild-like setting rather than a profit-maximizing setting, the masters are selected based on their skill in their craft. Their skill as a manager is not considered. This means that the intermediate and apprentice ranks are likely to be mismanaged by their masters a substantial proportion of the time. Since the intermediate and apprentice ranks are most of the people, they provide most of the labor, meaning that a substantial proportion of the labor is likely wasted.

Another inefficiency is in training. People imagine that the master would be a better trainer of apprentices than a intermediate rankee, because the master knows more. But actually, the person who has just recently learned material themselves is the best trainer; masters tend to forget what it was they didn't use to know, and hence tend to be bad at explaining things in a manner comprehensible to novices, as well as at devising a complete and balanced curriculum.

Besides the motivational benefits, a supposed strength of autonomous systems are innovation; if you didn't care about that, then it would be better to have a central authority tell the masters what to work on, to realize the synergies of having the various masters working on related things which are part of one big programme. But, in the fiefdom system, another inefficiency is in innovation. The apprentice immediately sees where the system could be improved, but for each valid idea they have five infeasible ones due to their lack of understanding. After they acquire understanding, they are still somewhat likely to be burning with enthusiasm to change the world and make their name; now is the time that they should be maximally empowered, immediately after apprenticeship. However, if the time from apprentice to master is overly long, then by the time the apprentice becomes master, they cannot remember their ideas for change; those ideas that they do have which rely on the intuitive understanding of youth of new technologies and new social systems are decades out of date; they are no longer burning with enthusiasm; and they have perhaps become enamored of a subspecialty that they have developed over the years, and loathe to throw that away to work on something new.

The end result is ironic; a system that used to be praised for its autonomy, in which individuals, once deemed competent, each chose for themselves which work to take, and each enjoyed the fruits of their own work, has devolved into one that is highly hierarchical, in which the vast majority of labor is done within fiefdoms, many of which are highly exploitative.

Many consulting partnerships (master = partner, apprentice = associate, intermediate = director), and some fields of academia (master = professor, apprentice = grad student, intermediate = postdoc), are examples of this antipattern.

The antidote to this antipattern is not to create the intermediate rank, and instead to elevate apprentices directly to masters. Be sure and make it possible for young masters to voluntarily choose to collaborate with the elder masters, in a manner similar to the old intermediate rank, if they wish to learn from them.

You might say, "But we can't afford to pay that many masters. There aren't enough available master positions." This is incorrect. Those intermediates are getting paid somehow in the status quo. You can afford to pay them the same amount they are being paid now, only also giving them the autonomy to choose/find their own work, and giving them the revenue/profit-sharing if any. It is okay if more senior levels of masters have a higher level of base compensation; the key is that all masters have the same level of autonomy in finding work.

Less radically, keep the intermediate position, but make it non-competitive, and make the criteria such that it can be achieved within a couple of years; so that what you are calling 'apprenticeship + intermediate' is really just equivalent to a slightly longer apprenticeship, possibly involving multiple masters.

Alternately, dispense with the fiction of autonomy, and create an institutional framework in which (a) the intermediates are not exploited as much as in the fiefdom model; do this by (i) having organization-wide processes to select, train, promote, and reward the apprentices and intermediates, rather than leaving everything up to their one master, (ii) paying the masters less and the intermediates more, and (b) the masters do not choose their own work, but rather are accountable to a central authority.

hierarchical inversion

When the person with 'natural authority' is in the employ of one without it, yet both are doing the same kind of work on the same task. Note that i do NOT mean the situation in which a manager employs a team of specialists -- in this case the manager is not doing the same kind of work as the specialist.

I mean a situation such as for example, when a consultant is hired to do a task while at the same time training a team of people to do that task in the future. The people will be best trained if they do as much of the task as possible themselves without relying on the consultant; but since they are the employers of the consultant, they will feel entitled to demand that the consultant does as much of the task as possible so that they don't have to; they will ask the consultant questions at every opportunity and not actually learn to do the task for themselves.

competitive committee approval for small innovation projects

A system in which innovative projects must compete for funding by making elaborate proposals to a distant committee.

There are four problems with this. First, the innovative personnel end up spending a significant fraction of their time proposing to work rather than working (both the proposers' and readers' time is wasted). Second, it seems to me that if 1 proposer has thought about the project a lot and think it's such good idea that they want to work on it personally, and if a committee of 5 somewhat more experienced people think it's a bad idea after reading about it a little, there's a good chance that it's a good idea (and even if there is only a small chance, if the organization is pursuing innovation it might want a portfolio of high-risk/high-reward projects). Third, when a proposal is rejected it will often be modified until it is accepted, after which basically the same project is done in the end; in which case the process was just a waste of time. Fourth, since the system is competitive, eventually people won't be able to propose their projects until after the project is halfway done (because at this time they can make a better case for it, having already explored most of the angles, and perhaps collected evidence showing that the project is useful -- that is, i'm saying that proposals which are based on half-finished projects will have a huge advantage over other proposals) -- this implies that they are spending half the funding earmarked for an approved proposal working on their next prototype, so the system isn't really controlling what they are working on anyways.

There's two good things about this system; first, in the course of thinking of how to explain their idea, the proposer(s) improve it; second, by forcing the committee members to take time to read all these ideas, information about what is being worked on and who is working on it spreads. But these don't outweigh the disadvantages.

The antidote is simply to empower a person who would be entrusted to sit on that committee to just fund a project by themselves. A more radical antidote is to pay innovative and competent people a salary, tell them to work on what they want, and then give them a bonus at the end if their project turns out to be useful.

I'm not talking about innovative projects which take hundreds of people to complete; obviously such a large expenditure of resources will require some approval. I'm talking about projects with about 1 to 9 people (one- or two-pizza teams). I'm also not talking about a situation where there is a clear roadmap, with a critical path, for some organizational goal, and the people in question need to be working on that roadmap, rather than expending effort on other projects.

underhiring of support personnel

Support personnel are cheaper than subject matter experts, so it's efficient to push as much work as possible onto the support personnel, and have the subject matter experts only spend time on things the support personnel aren't qualified to do, in order to get the maximium amount of work done per dollar.

When money is tight, it's tempting to have too few support personnel, since the experts can do everything. But this is actually a waste of money.

overhiring of support personnel in the face of 'neccessary complexity'

On the other hand, sometimes support personnel are hired to do tasks that really should be left with the experts.

For one example, there might be a pool of subject matter experts who do technical stuff and then a pool of client relationship managers who interface with clients. In theory, this allows the experts to focus on their expertise and the relationship managers to focus on people, but in many situations the client frequently needs to talk about technical issues directly with an expert; passing messages thru the relationship manager can become a game of telephone. In theory, this would work out fine if the relationship managers would quickly 'escalate' any request, and connect the client directly to the relevant expert, but in practice there may be pressure on the relationship manager to shield the experts from what they perceive as unnecessary requests. In this case the problem is that part of the 'necessary complexity' of the expert's job is communicating with clients about each client's unique technical needs, and it is not possible to offload this onto relationship managers.

Another example is hiring a secretary for someone who doesn't really need one. In this case, the boss may try to use the secretary as a 'shield' to keep annoying requests from other departments from reaching them; but (assuming the secretary was hired as support staff, rather than being an up-and-coming manager being mentored by the boss) the secretary will often be unable to (most commonly, not because they don't understand, but because they don't have the authority to make whatever decision needs to be made) actually answer these requests themself, meaning that all that is accomplished is delaying requests and insulating the manager from the needs of other parts of the organization. This is frustrating to all; the people making the requests will sometimes assume that the secretary is somehow managing their boss and has the power to make them pay attention to something that is important/urgent, when in fact the secretary has no such power. The secretaries may end up getting used as a punching bag, absorbing stresses between different components of the organization (ie instead of getting annoyed at Important Person A when they don't approve some budget on time, you can get annoyed at their secretary for making them do it, and be more congenial to Important Person A; this lets all of the higherups remain friends, but at the expense of never correcting Important Person A's behavior, and also of unnecessarily stressing out the poor secretary). In this case part of the problem is that part of the 'necessary complexity' of the manager's job is responding to various requests from other parts of the organization, and this cannot be offloaded onto a powerless secretary; hiring a powerless secretary for someone who doesn't really need one allows them to try to skirt this responsibility.

Another example is hiring someone to do a task such as recruitment when all they actually do is post job descriptions to a predetermined site and then forward all resumes received to the person requesting the hire. In this case, the result may be that the person doing the job, in trying to add value, will want to meet with the hiring managers to discuss how they can improve the wording of their job ad, etc, when the hiring manager may not want to spend the time to do that. In other words, hiring this extra person may create more meetings. In this case the problem is that many of the annoyances of recruiting (ie deciding what you need in the position, and where the type of candidates you need can be sourced from) are part of the 'necessary complexity' of the manager's job, and cannot be offloaded onto someone without subject matter expertise in the field that you are hiring for.

(hypothesis: some of these jobs work that way out of tradition, because in the recent past, before communications technologies such as Google Calendar, online meeting room reservation systems, online purchasing systems, online HR administrative systems etc, you had to either write a memo or get someone on the phone or schedule a meeting in order to get the smallest thing done, which means you have to followup or call back until you get them, etc, which added significant overhead to someone's day)

Hiring support personnel 'for everyone'

In the face of the previous antipattern (overhiring support personnel), it is tempting to say, well, there are SOME tasks that a support person could do for others, it's just that there's not enough of these tasks to saturate 40 hours per week of a support person for just one boss. So, let's hire a support person for the whole workgroup, not just for the boss.

The problem is that the support person might be disinclined to take orders from the other subordinates of their boss. Perhaps if the boss were explicit about this it might work, but more likely the higher-ups thought, 'okay, i guess we can justify hiring a support person, not just for this boss, but for this boss and their group', while the boss thought, 'i'm important and i should have a secretary, and perhaps the secretary will do some other things for other people too'. So, regardless of the higher-ups' ideas, the support person is really just working for the boss. (to test if this will happen; if the support person is lackadaisical about requests from non-bosses, but always does enthusiastic and excellent work for their boss, how will this get them fired?)

Centralized support pools face the same problem; see for instance in the comic Dilbert, the character Mordac, the Preventer of Information Services. The issue is that most of the people coming to the support staff will not have the clout to demand the solution they want; which in many cases means it's easier for them to just do things themselves. However, a centralized support pool is still a low-cost alternative that helps some people a little bit without having to pay for a larger number of embedded support people.

(i wonder if a PieShare?-based company could solve these problems? Or, another solution would be to give everyone, not just bosses, their own budget, and then have internal charges for support staff; allowing ordinary workers to be paying clients and get somewhat better service)