We conduct a character popularity poll both domestically and internationally
They have a pretty good idea of what people like.
We also research games that are currently being developed.
They try to include upcoming games to try and have a wider selection of characters age wise. Said characters still have to exist around the time the choices are being picked however.
After that, it's pretty much up to us, with a touch of convenience
The dev team picks the characters for the base roster.
There are a lot of things to take into account when deciding characters
This one's self explanatory. Why did I quote this?
Does the character have something inherently unique about them?
The dev team needs to know what the character will bring to the table. In my experience, the answer to "do they have something unique" is always yes, even if it's not a big sweeping mechanic, or even if it's another take on a mechanic that has already been tried.
Can that be used to make Smash Bros more interesting?
The meaning of this is a lot more up for debate. What do they mean by interesting? In terms of deeper gameplay? In terms of who's drawn to the game? Both? Neither?
Is the character a representative character [for the source material]?
The dev team wants to pick characters that can accurately represent the game/series in question. Generally, the first pick from a game/series is the protagonist since they represent how their game is played. It's also why a bunch of characters from veteran series (particularly from, but not exclusive to Fire Emblem) are focused on untapped aspects of their gameplay. Antagonists tend to emulate how it feels to fight the enemies in their games.
Do they conflict at all with other characters?
I don't really know what this means. Perhaps they don't want the characters to be too similar? It doesn't make sense to use the word "conflict" in that case. Are there mechanics that cause glitchy things to happen or are unfun when they interact? This makes a little more sense but I can't imagine that happens often if the mechanics are designed well. Unless the characters are hard to distinguish due to their designs (which likely doesn't come up often either) I highly doubt they're talking about visual similarities.
Was there anything about them that stood out when we began development?
My guess is that this is about tackling issues from a developmental perspective rather than a conceptual one, but if we're still conceptualizing characters at this state, this could be referring the idea of "finding the fun" where they play around with features until a direction speaks to them. In this case, maybe a Special Move does something unexpectedly neat when they first implement it, and they decide to lean into it harder.
Characters from a series that has no future are rarely chosen.
This is true to a decent extent, but it's in no way a hard rule, as we usually get about one character that fits this description per game for various reasons. Basically, it means that if your most wanted is from a dead series, you pretty much have to get everyone to shout their name from the rooftops and hope that's enough and/or hope the character speaks to Sakurai at some point during the selection process; They're far less likely, but they're by no means impossible (in fact I've probably painted a bleaker picture than what's called for).
It's worth noting that it's a lot easier to overcome this if you've already hit it big. See: Characters like Duck Hunt and (potentially) Simon.
Also, fighters that switch models are a different matter
All that can really be gathered by this statement is that swapping characters have a tweaked criteria. It's probably referring to characters like Sheik, Squirtle, and Ivysaur, who would not be here if it weren't for their connection to Zelda and the Pokémon Trainer respectively, but I can't be sure.
