Drifting to sleep, maybe asleep and resurfacing to wakefulness my mind was flitting around from idea to idea, from memory to memory. What I remember and was left fully awake with was two things: Queen's 'Don't Stop Me Now' stuck in my head and the notion that it ought to be possible to create a deeply detailed census of the entire world population at a point in the past, provided that point was recent enough to be reached by many genealogists, but not so recent that the world population was in the billions.
Now, the more recent a point, the more accurate the data and the greater the likelihood of living descendents, but also, the more daunting the project (due to the number of individuals described) and thus the less likely of enticing participants to join in the grand adventure.
As interested as I am personally in the year 1600, I know from my own genealogical and historical research that it is distant enough to be problematic. Jumping forward to 1750 would give an estimated world population of 700-825 million people. Or, by other estimates, of 629-961 million. That's a lot, but not an insane number of nodes. For example, using the former range, it's about the number of articles in Wikipedia in Chinese or in Portuguese.
1750 has also got the inspirational benefit of a big anniversary coming up within the probable lifetime of the participants or their children—300 years in 2050.
So, how to begin?
Infrastructure is vital. It must be incredibly robust and flexible. It must have profound internationalization support. It must allow for advancement and diversification separately of its data storage, software interfaces, and human interfaces.
Data will come in in many forms and must be clearly associated with its source, so that later conflicts on details can be weighed based on their respective supporting data.
Detail will vary wildly from broad guesses of total population in a country to general counts of categories of individuals (e.g., heads of household, taxpayers, members of the military) to detailed nodes about a specific person (both the famous and the genealogically derived).
Eventually, participants will no doubt be interested in assessing the relationships between individual nodes, thus it would be helpful to be able to retain data details (e.g., membership of an individual in a particular tracked category such as "the 12th regiment of Lord So-and-So's light horse", or "household at 123 Elm St, Anytown, New York, USA", or "inventory of the slave ship blah-de-blah", or "signatories of proclamation X".)
Such detail nodes will, of necessity, be much greater in number than the number of individuals alive because merger of them as applying to the same individual will be a more gradual and difficult process. This is a vital factor in infrastructure design.
It's used by some sources as a baseline year for the end of the pre-industrial era; rather nice as a stake in the ground for pushing back our knowledge of individual human participation.
The population of North America is only about 2 million, thus forcing U.S. participants to think about the world outside their borders (which I think is always a good thing). It also makes an enticing early goal for "near complete description", which is the best I'd expect we can hope for in any region.
Sweden begin taking a census in 1749, one of the very few countries doing so in the mid-18th century, and is thus a logical target for another "near complete description" goal. Conveniently, it's also a good country for online project participation with its highly tech-savvy population. The 1750 estimated populations of Sweden (which I'm presuming refers to its territory then, not its smaller borders now) 1.7 million or 1.78 million. (Pleasantly for me, it's also where I am pretty certain I have personal genealogical data for 1750. Been a while since I was working on my paternal grandmother's line, but I recall it going back that far and farther thanks to the good data there.)
Iceland is also promising for early population data and participation.
Now, what haven't I considered yet?
Thought which came to mind after I went back to bed:
Every part of this idea needs further definition, but particularly the area around what defines a counted individual. Chronological confirmation of someone with a citable source is a big part of it; that is, an individual for whom we have a specific record of them being born, dying, marrying, becoming a parent, or otherwise being specifically one of those alive at some point during the year 1750.
However, those records may actually be less evocative of human experience than the categoric description associated with what I'm calling, for lack of a better term, 'unmatched individual details', or 'unmadeets'. Whose story would you be most interested in, the confirmed individual "Mary Jane Smith born 1750, later the mother of Winifred Harding", or the unmadeet "one of 350 purchased slaves who rebelled on the ship King David at 5a.m. on May 8, 1750"? Which says more about what was going on in 1750?