Truth be told, I have seen some impressive "seamless" AI, the most obvious example coming to mind being Forza Horizon 3, where an AI model is trained to ghost your racing style and mannerisms and act as a human player without informing the other human players it's an AI. It's still blatantly clear that the AI is not human, but at least the game remains fun. I still have reservations on how palatable this might be to a board game-style game, but I can agree that a really well-made AI might be alright. Let's say Nintendo can make an AI that can suitably replace a human player, that still doesn't address issues of players dropping in the first place. It's then still an issue, even if remaining players don't notice, as you have to ask "Why are players dropping?" and a designer's role there would be to mitigate match drop rates.
What reasons would you give for people actively wanting to drop in and out of player positions? I think that's a patently false assumption. Why would a player wanting to sit down for an online match wish to inherit another player's personal progress? It's like sitting down to play Monopoly and taking over someone's hand. Not everyone is playing this game in a drunken stupor. A game designer who believed the majority of players of a board game would wish to start in the middle would be laughed out of a meeting room immediately, as it's an asinine assumption. Exception given to games that actively include that as a unique mechanic, of course. As I mention later, players are likely to drop an online MP game if they're losing and don't feel like they could win. Players who drop in are then most likely to inherit losing game progress. That's not at all fair or fun for the player.
Further, once again your idea of an inherited progress system... There are simply too many variables to account for to make such a seamless transition possible while keeping the game fair. Once again, would you go around merging games of Candyland or Chutes n' Ladders? My thinking is that players would be very unhappy with such a system. Just as people wouldn't take kindly to merging board games in real life, they'd be irked by having board games merge digitally. Let's also talk about the logistics of making that possible. It would require very hard balancing and a lot of iteration to get that general idea into a state that's passable as a fair online experience. I can't see a development team getting together and deciding that this is the appropriate way forward when a re-design of the core MP experience to better suit online would be a much more sensible option, and one more likely to actually succeed on schedule.
"If you don't like it then don't play it" is not an excuse to put in a broken mode. I believe MP as it exists now would be a broken mode in online for the reasons I stated earlier. You and I both know how fickle gamers are. It's easy to say now, "Just make it online and whatevs!" but imagine they do just that, and it has the exact problems I mentioned. Would people not then complain again? What designer worth their salt would release a game without appropriately re-designing elements that need it?
Let's consider some of the changes they could make. For one, they would likely need to quicken the game pace. What if, like Smash Tour in Smash 4, every player got to move at the same time? Or at least rolled at once and pre-selected a path simultaneously? Other tweaks, cuts, and additions can be implemented to quicken the game. Next, they'd have to make balancing changes in order to keep the game interesting for players who aren't in the lead. For a long game, players who are behind and who don't believe they can catch up are most likely to drop. If you're losing in a Mario Kart match, the race is only about 3-4 minutes so losing players are more likely to stick 'till the end. For a longer (and slower) game like Mario Party, people are likely to drop more quickly. "It's just a random game with random people, who cares if I drop right now?" Why would a designer not re-design parts of MP in order to make sure the least amount of people drop? Tweaking mechanics and adding "carrots" to keep even losing players interested and incentivized to see a match through to the end would be best. Couple that with quicker gameplay, and the problems I mentioned are mostly mitigated. The game would be more suited for online player, and if done correctly, would still feel like Mario Party.
Designing an online experience is not as simple as throwing your existing single player mode there unchanged. Some games lend themselves to being an online experience more than others, but even games like Mario Kart had to make major behind-the-scenes changes in order to provide a fun multiplayer experience. Even more work is required to do the same with Mario Party, however the net outcome ideally is an redesigned online Mario Party experience that still feels like Mario Party.
Unless your reason for the drop ins and drop outs is "because it is fun to troll people with it" (see various games where that would dump you back in matchmaking) then I would view that as something of a non issue and where that "if you don't like it" thing came in. Similarly with all this talk of AI I get the impression you view mario party as far more serious business than I, or indeed most, -- people still make smash brothers and pokemon into competitions despite both games being wildly unsuited to such a thing. It need not be some kind of game Turing test, just "good enough".
As far as AI problems go then Mario Party AI is perhaps not trivial but nothing I would expect any moderately competent AI team to be able to handle in normal game development timeframes. If you can make a passable collectible card game AI (
https://magic.wizards.com/en/articles/archive/duels-planeswalkers-all-about-ai-2014-02-13 ) then this is nothing compared to that. Outside of minigames (themselves no great shake for an AI) the number of variables and meaningful decisions in Mario Party is so low when all is said and done that it is no great issue. Go is a hard problem for board games, mario party board game section is not.
Other than disconnects/pauses then I am still not expecting players to drop into slots they in any way would perceive as inheriting someone else's position. It can work well in some games (people are more than happy to join companies in real life, and we already mentioned the likes of battlefield) but I agree it is less than ideal here. Equally it is still an optional and out there idea separate to the core concept I was heading towards.
With that said I might actually offer that as a quick play match option. Maybe throw in a catch up mechanic as well if it works (similar to double experience/rested player bonus in a mmo or something).
One might argue that snakes and ladders and candyland are not games (do you make any meaningful decisions in either, assuming we are talking about western snakes and ladders/chutes and ladders?) but that is a different discussion.
Variables wise I still reckon you could make it seamless. The chances of any two instances being in the same position at one point in time? Miniscule no doubt. You being able to pervert a couple of games in progress by holding or advancing AI players over the course of a few turns to reasonably match/sync and then merge? Far far more achievable. What is the spread of places you can move per turn? How many times have you seen a fairly radical redistribution of stars or whatever? Abuse that a bit and you can sync games no problem.
Is it fair in the traditional sense of the term? No, not even close. Keep it as something possible, possibly even plausible, in the game and would anybody likely perceive a problem? I would say no, see also mario kart rubber band AI.
"For a longer (and slower) game like Mario Party, people are likely to drop more quickly. "It's just a random game with random people, who cares if I drop right now?""
Not saying it is not a concept at play. I am saying who cares and that such a thing is an absolutely nonsense reason to not have what would technically be trivial internet play in a system crying out for such things and having to prove its worth at every turn.
Similarly some kind of karma system, possibly an invisible server side one, might be in play (it is fairly easy to tell disconnects vs poor internet if you do really care.
"players are likely to drop an online MP game if they're losing and don't feel like they could win"
I am familiar with the idea. Not sure it represents a game breaking issue here, indeed with instant replacement AI (possibly even instant replacement uber AI) it is even less of an issue than it is in real life games. I would possibly go a bit further and say it is the minigames that will keep people there -- one of the common solutions to said problem is to keep it such that the mechanics themselves provide the reward. If your minigames are not fun then there are bigger problems. Actually I might argue that the minigames are the reason for the game and the board section is an illusion of choice, some kind of cooldown/pacing mechanic, and a bit of light griefing. In which case the quick play inheriting option makes even more sense to me.
""If you don't like it then don't play it" is not an excuse to put in a broken mode. I believe MP as it exists now would be a broken mode in online for the reasons I stated earlier."
I am having trouble getting anywhere near it being so fundamentally broken as to make the exercise not worth it. It seems so obvious a move to put it online and the game seems so primed for it.
I imagine they could make a few changes to improve some aspects, such things are possible for any game out there, but I still see nothing that would fundamentally cause the experience to fail, not even close. To that end I find the resistance to the notion, and the criticism of the criticism of lacking it, to be utterly bizarre and incomprehensible. Even more so if they had friend code shared session play in it.