Issue No. 05 - May (2005 vol. 38)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MC.2005.167
Tired from the long Oregon to Boston flight, I was standing at the front desk of my hotel doing a routine check-in. Suddenly, the main doors burst open, and a small army of 15 very tall beavers swept into the lobby in a curiously synchronized fashion.
Upon closer inspection, I saw that the beavers were actually humans who had evidently survived Extreme Fur Makeovers. At the center of this group was a man with huge Elton John-ish sunglasses and a fur coat so thick that making it undoubtedly sacrificed an entire evolutionary chain of mammals. This guy turned out to be a moderately famous pop star, or so I was told by the hotel staff.
I had the eerie impression that "Elton" had some kind of strange attractor quality that was causing the members of his entourage to constantly reevaluate their physical distance from him. When that distance began to approach some upper limit, immediate corrective action was required, even if it meant displacing some other member of the swarm, who would then perform the same distance-plus-corrective-action algorithm.
The most striking thing about this group was their overt hierarchy. It was clear that "Elton"—the top dog, the alpha male, the big kahuna—was literally the center of mass of this group. I don't think he was exerting any effort to stay in the center, any more than protons have to work to stay in the middle of their atoms. The sunglass-wearing protons go where they go, and the cheap-fur-wearing electrons just have to keep up.
The effort was all being expended by the fluttering sycophants, who seemed to have a hierarchy of their own. Those closest to "Elton" stuck to him like glue, refusing to be displaced by anyone. I couldn't be sure, but it sometimes seemed as if these sycophants had their own sycophants. It's hard to tell since the swarm only had to flow (if you've ever seen one, you know swarms don't walk) about 25 meters from the street to the front desk.
Surreal as it was, this experience got me thinking about the hierarchies that develop when humans get together. Why do people seem to naturally want to sort themselves into a hierarchy or pecking order? Is it useful, or even necessary, or is it simply an artifact of our evolution?
It's clear why some organizations are explicitly, inarguably, unabashedly structured in an ordered fashion.
Military groups cannot operate successfully as a democracy, for example. A fundamental requirement of a democracy is that all of its voting members have the basic information needed to make intelligent decisions. Military groups must restrict the dissemination of information, partly for security reasons, but also for practical reasons: They almost never have complete information, and they may lack the real-time bandwidth to pass along what information is available. Beyond all that, there is sometimes an overriding need to take action, now, with no time to discuss things or collect votes.
The military takes this hierarchy very seriously, going to great pains to reflect it even in their uniforms. Soldiers can think whatever they want to about those above them in the pecking order, but refusal to abide by orders can have grave consequences.
As with many practices from the Far East, karate instruction is a stylized, formal affair. Before class, the students enter the practice room ("dojo") and stretch or practice their routines ("kata"). As the time approaches for the class to begin, the students line up along the wall and wait for the teacher, or "sensei," to enter. This is a gesture of respect from the students to the sensei, and it's a great way to remind them of their role and proper mindset in the class. His role having been acknowledged, the sensei doesn't need to prove himself in any way, and the class can then shift its focus to the students' progress in the art.
Conversely, is there anything worse than a person in charge of other people who is feeling insecure about his own authority or ability and doubts that his charges feel the necessary amount of respect for him—or worse, correctly perceives that they don't?
I once endured a college course in which the teacher was himself a recent graduate with no experience and had what turned out to be a justifiable lack of confidence in his ability to teach or even control the class.
Instead of starting the class in the usual way, by explaining what would be expected of us, this unfortunate soul chose to lecture us on how we were all born with silver spoons in our mouths and how it was incumbent on him not to coddle us even further.
Lacking the maturity to deal gracefully with this affront on the first day of class, and feeling our collective intellectual maturity had been questioned, our class gleefully chose to challenge his authority. From that day on, we often brought spoons from the cafeteria and solemnly placed them on our desks at the start of each class. The symbolism of this act wasn't lost on our hapless instructor, and the class remained rancorous for the rest of the semester.
Gauging the dynamics
In my experience, it is often the first-line supervisors, people new to the game of managing others, who are at highest risk of getting this dynamic wrong. Perhaps this is evolution in action: first-line supervisors who are not effective are less likely to be promoted further.
I've seen this in the Boy Scouts, for example. Most boys rise to the challenge of leadership in exemplary fashion. But as they move through the ranks, a few resort to using their new positional authority to subdue, harass, and intimidate their charges.
See the movies Lord of the Flies or Peter Pan and the Lost Boys to watch boys arranging themselves naturally into a hierarchy. Is there a name for the parallel sorting algorithm they seem to naturally employ?
In industry, newly promoted first-line supervisors must find a way to continue exerting influence on those who until recently were their peers—the same influence that got them promoted in the first place—without fatally disenfranchising their new reports. Achieving this new balance of power is fundamentally a matter of having all concerned accept the new implied pecking order.
Asserting positional authority is a poor way to accomplish this. We all know how easy it can be to give the appearance of accepting a new pecking order, while our heart isn't really in it, and expressing our real feelings by subtly undermining the new boss. Throwing your weight around is highly likely to elicit this subliminal sabotage. Such a situation isn't good for anyone.
Getting it right
When an electrical circuit is stimulated, there is an initial transient response, followed by a steady-state response. It seems to me that getting the new power hierarchy right requires a similar two-step process.
A newly promoted manager must first contend with the initial transient response by reassuring her charges that she's the same person they've known all along, that she won't grow horns and a tail just because she's now in their management chain (horns and tails are reserved for a much higher level of management), and that she realizes her own effectiveness is now inextricably bound to the productivity of the group itself. Over the longer term, she must also demonstrate the right steady-state response: She must succeed at her job. To do that, her group must also succeed.
Just after I was promoted to architecture manager at Intel, I got some very good advice from Albert Yu, former executive VP in charge of microprocessors. Albert pointed out that Intel's engineering ranks tended to abide by an informal, unauthorized, and unacknowledged pecking order. By some inscrutable consensus, this pecking order had the chip architects at the top, design engineers in the middle, and many other necessary but not always properly appreciated groups at the bottom—I'm purposely not naming them to avoid propagating these subliminal stereotypes.
Whether it was appropriate, justifiable, or even productive for this pecking order to exist wasn't the point. Because it was there, I would have to operate within it, and Albert explained that there were good and bad ways to do that.
A bad way would be to bully people and attend meetings with an "It's good to be king, kneel before me" attitude. Albert's proposal was that when you're in a leadership role, it costs you nothing to deflect credit to others whenever possible, and to explicitly include them in decisions. In effect, he said to never, ever reinforce their fear that deep down, you really do think you're smarter, or better, or more valuable to the company than they are—even if you do, in fact, think that. That kind of thinking isn't relevant, so don't let it intrude on the technical decisions to be made.
I think Albert had this exactly right—it defuses the pecking order emotional overlays that otherwise distort technical decision meetings and makes the process much more rational and linear.
Intel's 1990s chip development plan was often called a "waterfall" model: Design the best desktop microprocessor you can, and then modify as necessary to fit other markets such as mobile computing.
The executives in charge of the company's desktop business loved this model, but their counterparts in charge of mobile products thought they saw a few areas for improvement. As a partial antidote to what they feared was our desktop myopia, they assigned me to a corporate team to visit our mobile customers in Japan.
I had never been to Japan before, so I took a class or two on manners, customs and culture. I had read James Clavell's Shogun, so I thought I was already pretty well prepared, but no harm in going the extra mile. Konichiwa, anjin san.
One point made in this class was that Japanese society was overtly stratified, relative to the egalitarianism to which Americans aspire. We were told there were rituals to almost everything, from using elevators to taking cabs to exchanging business cards.
We practiced trading business cards and bowing appropriately. I even memorized the appropriate Japanese phrase for the occasion: "Hajimemashite. Yoroshiku, onegaishimasu." I'm sure my accent was atrocious, but I was hoping for partial credit just for trying. The person who taught me that phrase said he learned his Japanese from Tokyo bars, so I've sometimes wondered what, exactly, I was saying. "I'll have a martini, shaken not stirred" probably wouldn't impress my host.
It turned out that it didn't matter: I was so nervous and jet-lagged that when the time came, I only remembered my own name because it was written on the business card I was handing over.
Another thing we were told to watch for was the seating arrangements for business meetings. The Japanese company's highest-ranking representative would sit as close as possible to the center of the longest table. His underlings would array themselves in descending rank on both sides.
I smiled to myself when our Japanese counterparts proceeded to do exactly this. As with most high-tech firms, Intel prides itself on its "technical meritocracy," the idea that the only ranking that really matters is based on what you know and what you have accomplished. Overt shows of rank such as preferred seats in a meeting, corner offices with windows, and designated parking spots were deprecated as holdovers from a bygone era.
But my smug attitude evaporated when I noticed that our Intel contingent had somehow arrayed itself in precisely the same pattern: The highest-level VP had positioned himself directly across from his Japanese counterpart, and the rest of the Intel folks had seated themselves in the same descending order. (I was sitting off in the corner in the cheap seats.)
What were the odds that this was a statistical coincidence? I thought about the myriad meetings I'd attended in the US. Was there a pattern to their seating arrangements that I'd never noticed? Was I unconsciously obeying some cultural dictate or accidentally flouting cultural rules I hadn't known existed?
I concluded that, while an engineering team doesn't designate any particular seat in the room as the boss's chair, it was still true that wherever the boss happened to sit, the directly adjacent chairs were preferred over any others. We were, in effect, choosing seats based on their practical value (proximity to the boss increases interpersonal interaction with her and affords additional opportunities to impress her by muttering things like "Yeah, right, as if!" during the presentations.)
So maybe it wasn't a cultural dictum that set the Intel seating pattern, but it also wasn't the random "everybody's equal here" cultural value that we claimed to cherish. There was still a power hierarchy in evidence.
When you see a high-output engineering team in operation, look closer and you'll also see the ghostly but unmistakable outlines of a pecking order—several, in fact.
Teams that run smoothly do so because members know their own role as well as everyone else's. Tom is the go-to guy on questions about circuits; Karen is the tools guru; if you have a question about compilers, it's Maureen you're looking for. That team has worked out a complex web of relationships among themselves that is effective in both a technical and an interpersonal sense. Forging such a workable arrangement is crucial to development of a high-output team.
As with any other humans, engineers will arrange themselves in a pecking order. As their manager, your job is to be sensitive to people who are not comfortable with what they perceive to be their place in the order. Perhaps they have it wrong, and they are more highly regarded by their peers than they think (or vice versa, for that matter.) Maybe they aspire to greater things and need your help to chart a career path to that end.
Engineers like rationality, and pecking orders are only quasirational; there's a great deal of emotion and perception built into them. Be prepared to deal with the human as well as the technical.
If all else fails, tell them you studied karate, or wear one of those T-shirts emblazoned with "She Who Must Be Obeyed." Or try my one remaining bit of Japanese: Domo arigato, Mr. Roboto. (But beware: I'm not certain what it means, either.) Whatever you do, don't mention spoons.
Bob Colwell was Intel's chief IA32 architect through the Pentium II, III, and 4 microprocessors. He is now an independent consultant. Contact him at firstname.lastname@example.org.