Organizations fall roughly into five types:
Pyramids are an effective way to achieve repeatable results because they have the shortest path from the standard-setter to the worker.
They suffer from communication and supervisory faults because the organization is only as good as its weakest link. They lack creativity because they have poor communications ('why' is often is lost).
The classic fix for the communication problem is a magazine that reviews the whole hierarchy's business, perhaps daily or weekly. One good scheme has each person send e-mail up each week, telling what he did, his plans, and problems. Each boss makes a summary and sends it up. Then all the bosses send their summary down, appended to the summary from their boss.
At Printronix[?] this freed cash equal to a year's revenue, sped up engineering cycles six fold, reduced defects by two sigmas (see variance), increased inventory turns tenfold and doubled product service life. People found out what to fix, and where.
Hierarchies were satirised in The Peter Principle (1969), a book that introduced the term hierarchiology and the saying that "in a hierarchy every employee tends to rise to his level of incompetence".
These consist of a group of peers who decide as a group, perhaps by voting. The difference between a jury and a committee is that the members of the committee are usually assigned to perform or lead further actions after the group comes to a decision, whereas members of a jury ???.
Committees are often the most reliable way to make decisions. Condorcet's jury theorem proved that if the average member votes better than a roll of dice, then adding more members increases the number of majorities that can come to a correct vote (however correctness is defined). The problem is that if the average member is worse than a roll of dice, the committee's decisions grow worse, not better! Staffing is crucial.
Famously, unstructured committees can dither without making decisions. Parliamentary procedure, such as Robert's Rules of Order, helps prevent dithering.
A staff helps an expert get all his work done. To this end, a "chief of staff[?]" decides whether an assignment is routine or not. If it's routine, he assigns it to a staff member, who is a sort of junior expert. The chief of staff schedules the routine problems, and checks that they are completed.
If a problem is not routine, the chief of staff notices. He passes it to the expert, who solves the problem, and educates the staff -- converting the problem into a routine problem.
Staffs make decisions quickly, and carry out assignments efficiently, though less reliably than committees or matrices. For this reason businesses often prefer to use this method.
Staffs break down easily, usually from bad selection of people. Dilbert's boss is a non-expert trying to run a staff. In a "cross functional team," like an executive committee, the boss has to be a non-expert, because so many kinds of expertise are required. Also: chiefs of staff can be disorganized, play favorites, or can't tell what should go to the expert.
On the face of it, this is the perfect organisation. One hierarchy is "functional" and assures that each type of expert in the organization is well-trained, and measured by a boss who is super-expert in the same field. The other direction is "executive" and tries to get projects completed using the experts.
Matrices are the only known organizations that can consistently create complex technical products like airplanes and engines.
The problem is that going through channels takes too long. Getting approval to actually do anything often needs the approval of each type of expert, and both of each expert's bosses! The trick is to speed aprovals: make approval everybody's number one job, and simplify sign-offs.
This organization has competition. Bad parts of the organization starve. Good ones get more work. Everybody is paid for what they actually do, and runs a tiny business that has to show a profit, or they get canned. For example: upper managers invest, and if they make bad investments, there's no profit. Engineers rent their designs out to manufacturing. Facilities people rent space, etc.
This is a really effective organization. But it's wasteful because all those dead pieces of organization have valuable training, and are very hard to recycle. They're bitter, and they will stop taking it after a while. Reorganization follows.
These try to use each of the above types of organization in the right places. Very occasionally, a true organizational genius can make this work, for a while.
Don't bet on it in the long term. Success outgrows the ability of the genius. There just get to be too many special cases.
One golden exception may be a hierarchy of staffs, where every staff above the first level works to find or make the right people. This is the G.E. model, of course.
One of the latest areas of interest is Emergent organisations.