Coherence Penalty for Humans
michaelnygard.com
michaelnygard.com
"It will take one year," said the Master promptly.
"But we need this system immediately or even sooner! How long will it take if I assign ten programmers to it?"
The Master Programmer frowned. "In that case, it will take two years."
"And what if I assign a hundred programmers to it?"
The Master Programmer shrugged. "Then the design will never be completed," he said.
The topic came up on one of the introductory meetings and we shared our respective theories.
My take was that had this project had half the staff, it would take half the time to complete.
All and all there were over twenty people there doing daily standups.
There's also a related, from-the-trenches, story:
It is a very humbling experience to make a multimillion-dollar mistake, but it is also very memorable. I vividly recall the night we decided how to organize the actual writing of external specifications for OS/360. The manager of architecture, the manager of control program implementation, and I were threshing out the plan, schedule, and division of responsibilities.
The architecture manager had 10 good men. He asserted that they could write the specifications and do it right. It would take ten months, three more than the schedule allowed.
The control program manager had 150 men. He asserted that they could prepare the specifications, with the architecture team coordinating; it would be well-done and practical, and he could do it on schedule. Futhermore, if the architecture team did it, his 150 men would sit twiddling their thumbs for ten months.
To this the architecture manager responded that if I gave the control program team the responsibility, the result would not in fact be on time, but would also be three months late, and of much lower quality. I did, and it was. He was right on both counts. Moreover, the lack of conceptual integrity made the system far more costly to build and change, and I would estimate that it added a year to debugging time.
-- Frederick Brooks Jr., "The Mythical Man Month"
The general suspicion of TMMM is that it extrapolates from failure but assumes that some untried else would be better.
When someone makes three predictions, two turn out to be correct, and the third ended up not tried, we would do well to think twice before dismissing that third prediction.
Why are we taking advice from someone who lets others fail so he and his team look better? This kind of internal competition destroyed Motorola and damages many other organizations.
What makes you think he didn't?
> Why are we taking advice from someone who lets others fail so he and his team look better?
From the quote, Fred brooks was clearly the boss of both the architecture manager and the control program manager. He wasn't part of any one team. I don't see the kind of conflict of interest you're hinting at.
> It sounds like he was pleased to let them fail, so he could be correct with no effort.
To me, it sounds like he simply made a bad call, and was saying "oops".
Anybody who has worked for a couple of decades of large software project, doesn't need any, cause they have seen this play out time and again. Brooks doubly so -- he literally wrote the book on software development timelines.
Sure, technically you're right.
But it's not like every piece knowledge needs to come packaged in a fancy LaTeX, with confidence intervals, and control groups. Sometimes experience alone is enough to assure us that the rain is wet and that a team of 150 will invariably fail in this way when assigned such a task compared to a team of 10.
Where is the koan where the Master tells the manager to hire better programmers? The Master doesn't because it's bad advice up chase waterfalls, unicorns, and 10x devs.
"If it takes nine months for a woman to deliver a baby, getting nine women will not deliver a baby in one month."
I built a calculator that models that koan: https://www.andjosh.com/hard-numbers/
So maybe it is more story about how we fail when we think about organization and negotiation. (The master programmer making sure estimate so easily suggest they won't be at time anyway).
Another similar penalty with humans that is often ignored is caching of skills into working memory.
For example, consider a process, like writing an expense, that is really simple. Because it's simple, it might be tempting not have a specialized person do it and instead having every person do it on the need basis. But then, if it's done only rarely, people will have to learn it each time, or ask somebody how to do it, spending lot more time on it due to what are pretty much cache misses, and it would be more efficient to have a specialist do it, because then he would do it every day and had all the process corner cases in working memory.
Similar problem with caching happens when say a programmer multitasks on several different things at a time. In that case, the cache (working memory) is completely trashed every time a task is switched.
This is true, but it's important to remember that people aren't machines. Going too far down the division-of-labour route risks losing sight of what you're actually trying to achieve. Certainly in my case, that makes it a lot harder to perform my best, and it reduces the chance of spotting different ways to slice-and-dice the problem.