True if you need just one baby. If you need one baby per month on the other hand... Not quite sure how that translates to software development though...
Even then you would not get the first baby for at least 9 months. After that you would get one baby every month circa, but just as with most software development situations you can't speed up the first baby growth by throwing more people at it.
I've actually always disliked this analogy as it relates to project delivery (whether it be software development or a consulting project).
It suggests that there is nothing anyone can do to decrease the amount of time that a baby gestates; but that's quite different from project delivery.
While adding bodies does not necessarily divide the length of time required, there are absolutely concrete things which can be done to reduce the timeframe.
You can reduce scope (or features) for starters.
A woman doesn't have the option of delivering a "minimum viable product" of a baby (with only core features) in less than 9 months, but a consulting engagement could certainly cut out features to be completed sooner.
Childbirth is a subset of project delivery whose completion dates are not particularly fungible. There are numerous other types that are.
Part of my career has been doing software project management, and it's been very rare that an end client that I was working for was willing to extend a deadline or limit the functionality of a project. If a company is run and backed by software professionals this may happen, but for most other projects you just aren't allowed to change the rules — which is why the baby analogy holds so true.
I understand that the reason the analogy has become popular is because at some point, an idiot manager looked at a project plan, saw that the project was supposed to take (let's say) 6 months, and said "Hey, if we increase the number of developers from 2 to 6, that means it will only take 2 months."
So I get why it's easy to respond with "9 women can't make a baby in 1 month."
But let's say that instead the project plan said that it was going to take 6 months, and there was 1 developer on the project. It's absolutely likely that adding an additional number of developers can shave some time off that project (not inversely proportional to the number of developers you add, and not in every single circumstance, but I'd argue for the overwhelming majority of projects).
The analogy suggests that no part of project development can be parallelized, which to me just seems like shitty management of the project.
The problem with the 6 months 1 programmer vs 1 month 6 programmers is that the minute you have more than one programmer they have to have an architecture that works together plus take time to communicate with each other — and then add to that the concept of a "critical path" i.e. that one programmer may not be able to start something until another programmer is done.
Although in my professional experience the idea of adding more programmers is always suggested by the client about a week before the project is due!!!
PS If you had to live through about ten of these clients you'd have the same religious feelings that I do for the Mythical Man Month.
Reproduction is an interesting analogy for software development, but like all metaphors it eventually fails.
For instance there is the change of feature set as pointed out by m0nastic. Large projects tend to be very flexible in features over their development life, which is odd because usually its not the end user requirements that are changing.
Another thing is varying programmer qualities. Some programmers can bang out a baby in 9 days. Some require 90 months to make the baby (hint: don't whinge about your hiring praxctices, man up and get rid of these ones). Some are completely infertile (fire these ones with extreme prejudice). Some project a field of radiation that will not only prevent themselves from making the baby, but will kill all babies in a 5 mile radius (do the world a favour: take these ones out back and shoot them, and then burn the body - do not promote them to management!).
Continuing the reproduction metaphor, Yak Shaving can be considered similar to Onanism. While some people might not be able to make babies, they might be master craftsmen when it comes to building cribs. Some people might be bad programmers but really good at changing nappies... without these people you will be buried in a mountain of... 'cruft'. :D
Most activities can only be made partly parallel. The rest must be serial; that is, will depend on steps done before, and cannot be made faster by breaking it up for more workers. If a task is 90% parallelizable, and you multiply the number of workers by 9, you only get a speedup of 5. If you multiply the number of workers by 100... you still only get about 9 times faster. :-)
Oh yeah, I realize that, but the timing of it just seemed too close. That comment really summed up that thread... so anyhow, when I saw this article that was the first thing I thought of.
As a non-programmer The Mythical Man Month was perhaps one of the best books I've ever read to understand how to manage programmers, and how to explain to non-programmers how things work.
If your goal is feature poor and bug filled software then you can ignore this advice — but if you want to manage a large scale project with multiple programmers this advice still works today.