As I procrastinate doing something important, let's go back through time and figure out how to Fix Sega Forever and keep blue sky in games as a console choice, not just a software choice.
Presume that we can't move events forward (like launching the saturn in 1992) but we can move events back, cancel things, etc.
Here's my proposed first pass, which I'll split into eras/consoles:
SG-1000 (1983)/SMS(1985) era: no change. I'm content with how this went. Popular enough in the UK to keep the company going.
Mega drive (1988): My only suggested change is a different advertising approach in Japan to make the console more popular there. emphasize the RPGs, make a few more shooters, and maybe even appeal to national pride by talking about how the games from japan are so hardcore, etc.
Game Gear (1990): No change for the first couple years, EXCEPT I would launch close to simultaneously in the west and Japan. Then let tec toy do R&D and release a 2nd model in 1993-4 with a smaller form factor and more efficient battery usage, which you can then bring to the US, EU, and Japan. Translate the Last Bible games and get some pokemon stuff going with Game Gear.
Sega CD (1991): I'm torn here. I think the add-on worked but I almost feel like launching it AS the CDX or X-eye, an all-in one, might have been better. I'd also maybe delay the release to 1992 since in retrospect it wasn't that critical for the company to release this quickly, since the CD market wasn't crowded. Also here I would have sega document its processes better for developers, make better libraries faster for devs to use, and generally make the Sega CD as easy to develop for, and critically to exploit the new features of, as the pc engine cd was. More usage of the sound chip, more usage of the superscaler tech, less emphasis on FMV after the initial couple successes.
32X (1994): Here's where we get into interesting territory. I wouldn't actually scrap it, I would launch the 32X as a new console with Sega CD and Mega drive integrated, again with really nice documentation and tools, and give the genesis another two years of life. We'd have more 32x/32X CD games going into 1996/1997, but more "regular" mega drive games too. At the very lease release the Neptune instead of the 32X add-on.
Nomad (1995): Ditch the Nomad entirely and make a new handheld based on genesis architecture but with transparency support. Release this two years after the Saturn, once our long-running game gear has fizzled a bit, which will also be two years before the GBA. This lets devs make "enhanced ports" of genesis games to this new handheld console and let it communicate with the Saturn and Model 2 games. Genesis devs can continue to make games with tech they're familiar with too. This continues through the Dreamcast era.
Saturn (1994): Controversial again, but move the launch to late 1995 and base the system on the sega model 2 board architecture from the start, making model 2 arcade ports easy. Lose a bit of money on initial hardware sales but get the arcade scene totally on board with what you're doing. Create extensive tools and documentation for developers ensuring it's really easy to work on. More corporate espionage at Sony, and also: take Sony super seriously. I want nice 2D in here still but I'm not sure how to achieve that simultaneously without spending more money, in this era.
Dreamcast (1998): No change, though I'd also launch worldwide from the get-go! OR wait until 1999 to launch in all regions but with a slightly higher polygon-pushing processor. And now instead of trying to right the ship after the Saturn fails in the West, hopefully we now are building on its success.
In this way we hopefully get to a Dreamcast 2, and a handheld with the power of the Dreamcast that is basically a big VMU.
WELL those are my opinions, what have y'all got.