This is often attributed to Fred Brooks in his seminal book "The Mythical Man Month"

* https://en.wikipedia.org/wiki/Brooks%27s_law * https://en.wikiquote.org/wiki/Fred_Brooks

It also appears to originate from Theodore von Kármán (1957): "Everyone knows it takes a woman nine months to have a baby. But you Americans think if you get nine women pregnant, you can have a baby in a month."

I first heard it from Fred Brooks in Mythical Man Month. But it's even funnier if it came from Theodore von Karman. A lot of HNers will know him from KSP as the namesake of the Karman line, the boundary between atmosphere and space, typically set at 100km above mean sea level.

If you assemble a 9+ stage pipeline, you may reach a point where you produce approximately one baby each month.

Even then they are "working" independently. It's not a pipeline that passes the baby from one to the next. That's just plain parallelism. Either way it would do nothing to latency and so would not get the current one done any sooner.

That's not a 9 stage pipeline, it's 9 separate single stage pipelines.

Massively parallel Baby Processing Units (BPUs)

If you implement QBPUs (Quantum Baby Processing Units), you could theoretically produce infinite babies. Of course, we're perpetually 10 years away from working QBPUs.

That's throughput, not latency.

Most companies have more things they want to do then they can do in the next 6 months. So if you increase throughput without latency, you can still get more of those projects done. Even if you're still doing them all inefficiently.

It's ugly, but "don't hire more people because we'll be less efficient" often runs into this ugly "but we need more total output, not more efficiency" wall.

Yes, that's exactly the point of pipelining.

Sweet. The baby-in-a-month problem is reduced to assembling and priming a nine-stage pipeline in a month!