Yu Suzuki said pretty much this in an interview at some point. I can't find the interview now but he basically said that because they knew they weren't likely gonna make the third game for a while, he wanted to give players the chance to see Guilin and also slowly be introduced to fantastical elements (e.g. floating sword).
EDIT: I found the interview(s):
https://www.eurogamer.net/articles/...more-money-yu-suzuki-on-the-return-of-shenmue
http://www.shenmuedojo.net/forum/viewtopic.php?t=48967
My takeaway from this is that Shenmue 3 was never truly worked on beyond what might have been created for the Saturn.
The development seems to have gone like this:
- Shenmue is developed for the Saturn. This is a very embryonic form of what we got, and was likely a very different, very linear sort of game. I could be wrong, because I've not read much about this, but I feel it unlikely that it would have had the world clock, individual NPCs with their own schedules, dynamically changing weather/time, any of that stuff. There's no indication of any of it existing in the short video for it, anyway.
- Development then moves onto the Dreamcast, probably for a number of reasons.
- The games we know of as Shenmue I and II are developed as a single release. The engine is likely built from the ground-up from this point, to cater to the things I mentioned above.
- Halfway through production, it's understood that the project is too large and releasing it as a single volume makes little financial sense. Development refocuses on polishing and extending the Yokosuka chapter.
- Shenmue is released. Sales are underwhelming. The Dreamcast is failing and it's understood that the series will not complete on it.
- Development switches back to what we know of as Shenmue II, which would again be polished and extended. While this work is occurring, the entire future of Sega is in question. With that in mind, Yu Suzuki adds the short Guilin section we see in Shenmue II.
- Shenmue II has a torturous release, split across regions and consoles. The sales do not excite the beancounters.
- However, since Sega has already poured so much into this project, and since so much additional work has already been carried out (composed music, scenario planning, script-writing, environment and character conceptualisations etc.), they investigate further developing the series on the X-Box.
This last bit is key. I believe that money men in Sega signed off on spending such an absurd amount only on the promise of an episodic, serialised game that would re-use the same (expensive) engine and assets to hit a 1-2 year release cadence.
I think they did start very early planning on Shenmue III, if only to recoup their damn expenses, but quickly realised two things:
1. They would be reliant on people buying Shenmue III even if they hadn't played I and II (because of how those games came out). Someone - I bet you - someone would have suggested that they would see fewer and fewer sales for each episode of the series.
2. They realised that they would have to tweak the engine much more than they had between 1 and 2. So Sega would have had to sink another large chunk of change into a project that was already losing them money, at a time when the whole company was under existential threat.
Looking at those two things, they would have cancelled the project before it progressed past planning.
Of course I'm making a lot of assumptions here.
I want to expand on point 2, since it's shaky. I think the engine and assets of Shenmue were optimised for urban, gritty environments. The lighting was designed to simulate indirect light and indoor light primarily (Shenmue's cutscenes are defined, to me, by the constant framing of people's faces in half shade). I think you can see the effect of this on the Guilin chapter which, I feel, doesn't quite live up to its concept. The greenery isn't lush. The natural world doesn't look particularly natural. There isn't the hard transition between the shadows of trees/branches and the harsh direct sunlight that defines a walk in the forest. Most importantly, it feels like you've gone from walking through a bunch of narrow corridors in Kowloon, to walking through a similar set of corridors in Guilin, it's just that the walls have been replaced with trees.
I can't imagine this is what Yu had planned for Shenmue III, and I think this is borne out by how lush everything looks in the real Shenmue III that we have today. I know some people are put off by how dramatic the change is.
I think though, that we wouldn't see beautiful natural environments that would match the realism of the urban environments of Shenmue until Far Cry in 2004.
I hope Yu Suzuki will talk more about this after Shenmue III is released though. I think that the Shenmue III that we're getting is going to be dramatically different from the Shenmue III that Yu had in mind in 1999. Not just because of the technological advancements but because there is an expectation that must be met. That's why Yu's talking about forklifts and making the world larger and larger and more open etc.
There's no way that, after so many years, that Yu can stick to the "expand inward" idea. That kind of experimentation is only possible in an episodic framework where you can make one "episode" dramatically different without losing your audience - they've only waited a year or two for this episode, and will only have to wait another year or two for the next.
I do have an impression of what it might have been like (Majora's Mask on steroids!), however, and I still would love to play THAT game.