and you absolutely cannot follow a Sim anywhere – once they’ve entered a building, whether residential, commercial or industrial, the game stops following them, and good luck finding them after.
This made me laugh sadly. I remember playing SimTower when it was released in 1994, and you could follow any Sim around as he went about his day. It's now almost 20 years later, with vastly more powerful computers, and the AI engine is actually worse.
Rollercoaster Tycoon has the same mechanic—not only was each person a discrete, persistent individual, they had their own unique mood, stamina, hunger, tolerance for waiting, and individual tastes in amusements, food, etc.
To be fair, I'd be quite disorientated too if a huge magic hand came down from the sky and transported me to another location. ;-)
In all seriousness, I can forgive that kind of bug because it's the players manipulating the game in a non-realistic manner. If you don't push the sims about manually then mechanics were quite well done. In this, however, we're talking basic AI failing which is central to the game.
You're comparing apples and oranges. I'm sure the simulation engine in Sim Tower was wildly less complicated than in Sim City 5, or any other Sim City game after the original, for that matter.
Honestly, I don't really care about following individual Sims in a Sim City game. It's out of scope. If anything, they went too far in that direction in the current Sim City game, wasting resources on individual entities as opposed to having a solid macro model. I assume that's why we ended up with these pathetic city sizes.
They each have multiple methods of transportation. SimTower has stairs, escalators, elevators, and express elevators. You can see congestion as your sims' silhouettes pile up in front of elevators. To get to their destination, they can switch elevators once, and take four staircases/escalators. Sure everything is orthogonal and not curvy like SimCity5, but cities could be build that way too.
I don't have any special insight into the internals of both games, so I can't say for sure. I certainly think you could make a tower management game that has a simulation engine that is just as complicated as SimCity 4 or 5. On the other hand, I think that in principle a game attempting to simulate, to a degree, an entire city is bound to offer more complexity than a game attempting to simulate the goings-on in a single building.
I can make an argument from observation: Sim Tower is a 68030/386-era game, and from what I remember, my 68040 ran it very well. SimCity 4, OTOH, can still bring (a single core of) a modern CPU to its knees. Of course it's possible that SimTower was incredibly efficient and SC4 incredibly inefficient, but it seems much more likely that SimCity 4 is in fact vastly more complicated.
Simtower definitely seemed to simulate individuals and had a population cap of 15-20k. Its path finding was drastically simpler though.
Every floor (110 floors max) was essentially a single zone that could be crossed instantaneously so each person needed a path only from floor A to floor B. There were strict limits as well about how many transfers were allowed per path and how many total stairs, escalates, and elevators were build able.
Sim City 4 allowed many millions of "people" on networks of significant complexity. The SimTower simulation was simple, satisfying, and relatively bug free. It also wasn't complex enough to continue to be satisfying for 10+ years the way SC4 has been.
I wonder about relatively long-lived organizations like Maxis. Do they hold on to old code, or does it just disappear or fade away into obscurity? All of those games, most still playable, and all of that code have to present some value. If two decades later the performance of their new games is worse, however, even in this narrow area, it has to be that they have ignored their past.
Is it the right thing just to throw that old stuff away, or to ignore it? Is there any value in looking back at old successes, and even in reusing some of that technology? The people who wrote that code are probably long gone, but that kind of situation doesn't stop big old companies in other industries from still relying on their ancient COBOL systems, for instance.
Age of Empires 2 came out in 1999, developed by Ensemble Studios. Age of Empires Online was released in 2011, by Robot Entertainment and Gas Powered Games. A good portion of its code came from Age of Empires 2[1], including its terrible path finding[2]. Modern RTS games actually have pretty good path finding, and even Starcraft (1998) had better path finding than the legacy code used in AoE Online (2011). Microsoft published both games, and presumably they dictated the reuse of the old code. Probably to save money.
The Tropico series also do this pretty neatly (especially 3 and 4) - you can follow your 'tropicans' from their family home to the job, to the market for food, doctor, church etc. A policeman would follow your orders and run after someone you ordered arrested. Traffic path-finding is not perfect though.
But in Tropico, it is rare to have more than a thousand agents at any one time - whereas I think Sim City '13 has to simulate at least a couple orders of magnitude more agents. Not sure about that - I didn't buy the game.
I didn't buy the game either, but I've been following the news on it. What I've been seeing is that it's difficult to get population north of 200,000, though I would imagine not impossible as the accumulated skill of the player-base increases.
So, yeah, a couple of orders of magnitude... except, somewhat upthread from here someone makes the claim that a population of 100,000 sims actually corresponds to about 10,000 modeled sims, and then 90,000 "phantom population" on top of that. If true, this suggests only one order of magnitude more population than your Tropico examples, and the achievement is considerably less impressive.
This made me laugh sadly. I remember playing SimTower when it was released in 1994, and you could follow any Sim around as he went about his day. It's now almost 20 years later, with vastly more powerful computers, and the AI engine is actually worse.