What's so different about Blazblue updates?
Suppose you've baited your opponent into doing something really stupid or risky like mashing uppercut. You go to punish them. Here is what Ryu does in each iteration of Street Fighter 4.
Here is a set of bog standard combos Ragna used in the first Blazblue iteration.
And here is a set of his basic combos in the current.
Blazblue loves to completely rework its combo theory from the ground up each time. A lot of knowledge you've gained about what to do in certain situations, what's a good damage option vs one that carries further and so on get thrown out the window. The same goes for characters' tools in situations that have nothing to do with combos. In one game Jin can cancel his far-reaching 6C into a dash to do pressure with it. In another he loses it and Ragna gains it instead. One day Nu is a long-range character. The next day she's a close-range pressure character who happens to have projectiles. The next day she's a stance character. The next day she has the fastest standing overhead in the game. Don't get me started on how inconsistent they are with Tager. One day you don't want to hitconfirm with jabs because they'll scale the combo too much. The next they do nothing at all and may well be an optimal way to start combos. The next day they'll go back to reducing combo length, but only on some characters.
Whenever you get into Blazblue, the next iteration to come out might as well be a new game that happens to have some similar base system mechanics. That's assuming they didn't rework some of those as well.
Why do they do this?
Arc System Works are fiercely dedicated to keeping Japanese arcades running. Their entire business model for fighting games is built around keeping money flowing into them as much as possible. Their current process for games releases is this:
1. Release a game in the arcades.
2. Maybe patch it if something's really busted (hello there Chie 5B causing fatal counter).
3. Release it on consoles a somewhere between 8 months and a year later.
4. Release a new game in the arcades at the same time.
5. Do the console release for that game.
6. Announce new update of the previous game. Release it at arcades.
It works excellently. There's always something fairly new to play in the arcades that you just can't experience at home. That doesn't quite explain why Blazblue revisions work the way they do in comparison to, say, Guilty Gear's though. So there must be another reason.
While GG and BB are both by ASW, they have different directors. Guilty Gear's early iterations were headed by Daisuke Ishiwatari, and taken over by lead programmer Pachi for Slash through to Accent Core +R. Both worked together on Xrd. Blazblue is headed by Toshimichi Mori, who is known to ignore others' feedback. So we can conclude that he has a certain business approach to how to keep arcades making money that differs a bit to the others at ASW.
You see, if people have to relearn a character's combos in a game that is only playable in arcades, it means they have to spend more money. Not only will they be spending their usual allotment of yen to play with others, but they need the extra lab time just to get their skill at the level they were two months before the new iteration came out. As well as giving a game a sense of new life and a more even playing field for newcomers, they're double dipping on the existing userbase. It makes perfect sense from a business standpoint.
The game keeps trying to reinvent itself, so I think Mori may not be happy with the game's core. If a game needs to keep reinventing itself, it may well not be a very good game in the first place.
Players of other fighters, particularly the Capcom games tend to view airdash players as being a fickle bunch who jump from game to game as soon as a new one appears. Maybe it's because relearning Blazblue is in itself the same as learning any other game, so they're just trying to find one that's enjoyable enough to really stick with for ages. After all, the Melty Blood community stuck with its game from ReACT all the way through. It's just a shame Type-MOON tried so damn hard to prevent them doing so. That's for another time though.