Anyways, back to peace. My usual way of managing the specialists was to run maximum civil engineers until the factory was built. Then run policemen, since one fewer wasted shield turns into two with factory and Hoover and you get economic benefit as well, until all of a city's economic buildings were built. Then convert them all to scientists and push research. That last breakpoint happened in many cities as commercial docks were set up to complete as I got Mass Production, and by briefly making almost every specialist in the empire a scientist, I shaved Motorized Transportation research down to 4 turns total.
Happiness wasn't a problem with a decent supply of luxuries and both medieval church wonders. Unit costs were also not a problem as I was comfortably under the Monarchy free-unit limit. With policemen specialists eliminating corruption in most of my cities, there was no need to even consider Communism. One fairly big problem was pollution, but thank Sidness for workers on Shift-A.
Here's peeks into some of my cities, for how the specialists were getting managed. Augustodurum has three policemen to eliminate all its corruption, then scientists all the way.
Recently-captured Hieraconpolis is trying to get up to speed on buildings. It's maxed at size 12 so can put all the population towards civil engineering instead. Those civil engineers tripled the speed for the city to build the factory.
Richborough is building an important wonder, so after the one policeman to eliminate its corruption, everyone else is a civil engineer rather than scientists.
And here's an overall look again, with the island in a little inset image.
With the overall lack of shields, there was none of the usual military building during late-industrial infrastructure downtime. My cities went straight from commercial docks and police stations to prebuilds for research labs and offshore platforms. Yes, offshore platforms - that was my first research target in the modern age as those would be a serious boost to shield production.
Global Warming knocked out the forest at Entremont, costing me a few shields. It had a bonus grassland underneath it all this time, though.
When I got into the Modern Age, the AIs had run out of cash to pay me for anything, so I had to cut back from 100% research to 80%. Miniaturization took 10 turns to research. Then I researched Ecology; I really do need mass transits for these mass quantities of people. Global warming already killed one forest and I'd like to avoid losing more...
I spotted a Roman ship on what looked like a sneak-attack vector towards me. I moved a few ships to hopefully block it from its destination.
A few turns later, Rome declared war on the Portuguese. I don't know if my little blockade there actually dissuaded them.
Nope. Well, since Rome is also at war with Portugal, and Portugal is weaker than me, any alliances the Romans might buy would go against Portugal, not me. So I saw no need to buy alliances, and made peace with Rome after no more battles.
With offshore platforms, I could get several cities up to 60 or so shield production, which was plenty enough to prebuild the spaceship parts as the techs came in. The AIs declared a couple more wars against each other, with both Portugal and Carthage getting wiped out, but none against me. I got research on several techs - Space Flight, Synth Fibers, Satellites - down to 5 turns each.
No more sci GLs during the last two ages, but I could hardly complain about that after getting two in the middle ages. Two turns before Robotics came in, my Aluminum depleted, making me jump for a second, but then it reappeared in exactly the same spot. 
I actually worked out a method not unlike Sirian's "PebCat" for GalCiv to figure out how much research I could get from scientists. More specifically, this was to work out whether the time to research a tech could be shaved a turn from what the slider says.
First, set the slider to 0%, and note the number of turns for research (must be less than 50 for this to work), then do the same thing at 10%. Take the difference of that, which in this example is 37 - 24 = 13 turns.
Multiply your science production at 10% by the turns at 10%. 117 * 24 = 2808. This is the number of beakers you need for the tech excluding 24 turns of scientist production, which doesn't show up on the economy screen.
Divide that value by the difference in turns. 2808 / 13 = 216. This is the number of beakers per turn being produced by all your scientists. 216 beakers would be 72 scientists total empire-wide (divide the beakers by 3), which sounds about right.
Take that value and multiply it by the number of turns needed at 0% science. 216 * 37 = 7992. This is the true beaker cost that you still need to accumulate to get the tech.
At 100% science, the screen says we'll have the tech in 6 turns. Can we shave that to 5? It would require 7992 / 5 = 1599 beakers per turn. We're currently producing 1243 + 216 (scientists) = 1459 per turn.
We need 140 more science per turn, or 47 more scientists than I have right now. It might be possible with fanatical micromanagement and heavy worker merging, but it's not worth the effort. If it turned out to be that I was like 10 beakers short, I'd do it of course, and had done so for a couple earlier techs using this method.
Anyways, enough numbers. I win.
Not all that far from an 80k (small map) cultural victory, in fact. 3882 game points.
So now that playing this variant has forced heavy use of the specialists, what insights do we get for using them in a standard game? Well, even in this extreme a variant, they really aren't that useful before hospitals, policemen, and civil engineers.
But in the industrial era and late game, they're definitely useful. First, it is always correct now that a city with railroads/factory/powerplant/hospital should be irrigated and grown beyond size 20 until there are enough citizens to turn into policemen specialists to eliminate all the city's shield waste. Mining a grassland gives you 2 shields with a railroad; but irrigating it instead and making the extra citizen a policeman gives you the same 2 shields because of factory multiplication, plus one commerce income that can also be multiplied. Net result is the same shield benefit, plus economy.
Beyond policemen, you have a serious choice to make between mining for raw shields, or irrigating to make scientists for more economy that can also be swapped to engineers in a pinch. In any government, including Republic or Democracy, the fastest path to space would be to irrigate to the max, hire policemen as useful, then all scientists, swapping them to engineers as necessary for wonder and spaceship building. Irrigating and hiring scientists would often result in a few turns faster to diplomacy or space than would leveling off all the cities at size 20.
On a less analytical level, the industrial-era specialists are indeed quite useful in getting new captured cities up to speed. Civil engineers build the factory, policemen take over to get a courthouse built and almost eliminate the city's corruption, and you can build anything you want from there.
All in all, the specialists are pretty useful. Especially the policemen, which are borderline overpowered; with the standard multiplier buildings in place, they're like a civil engineer PLUS a taxman or scientist.
Thanks to everyone who played, and to Griselda and Charis for their consultation and advice in setting up the game!
