I find my play with five ends up very similar to what you describe with three players, except instead of two sets of activity (your group of two and then the third player alone) you end up with at most three sets (two groups of two and a lone actor) and often two (three and two or four and one).
I find the game makes some difference - my players will look for aid and back-up in Burning Wheel more readily than in Apocalypse Word, which I think is a function of AW characters being generally more badass from the outset than three lifepath BW characters. But even in AW it's extremely rare for me to have five players doing independent and unconnected things.
If I've made a move and Sickboy is missing, there's always more than one player that wants to know how and why, and that will naturally lead the players to seek each other out as different moves catch or require their attention.
I really like running for three in Apocalypse World, but it only takes two or three rolls to go against the players and things can spiral out of control for a hardhold pretty quickly. So I find with three, my players won't seek out confrontation with each other directly or overtly - there's enough adversity (and tension and fun) from scarcity and the apocalypse without really needing to.
Whereas, with five players you've got enough manpower to make the hardhold feel reasonably secure, and that gives greater opportunity for conflicts of interest between the characters.
Just a few quick comments on number of players in these games. As you and @pemerton
know (or at least I'm fairly sure I've expressed it to you in the past), I'm definitely a "3 is the sweet spot for players and 2 is far preferable to 4" kind of GM.
However, I concur with your above regarding Apocalypse World. The 10 month game I ran for it featured 4 players and there were sessions where I only had 2 players. The difference in interpersonal conflict that would occur between 4 players vs 2 was stark.
The games I've run with 4 players the last little while are the two long-running Blades in the Dark and Stonetop games I'm running. Homefront phase in Stonetop always seems to be around 2ish Threats/Opportunities regardless of player numbers. Expedition/Adventure-phase is more sensitive to numbers and/or magnitude. When only 3 players are present OR the magnitude of the Threat/Expedition/Adventure is large, its all-hands-on-deck. When there are 4 players present, its pretty much exclusively 2 players pursue one Threat/Opportunity while the other 2 pursue another (with both possibly, or likely, Requisitioning Assets/Followers from the steading). When it comes to Blades, 4 players is going to equal 2 Scores (sometimes 3) except in the most extreme magnitude of cases. My guess is, from the start of this particular Blades in the Dark game (correct me if I'm wrong @AbdulAlhazred
), we've only seen about 6 x Scores with all 4 members of the Crew on them at once; Cock-fight Assassination, Stazia's Stash and Grinder Warehouse/Armored Carriage Sabotage/Robbery, Lean on the Duskvol Times, Deathlands Perilous Journey, Assault on the Sun King Cult Compound, Assault on Lord Scurlock's Manor? Might be missing one (I don't think the Kill Bill-esque Bath-house Duel had Beaker).
Regardless, its absolutely doable to run Story Now games (including those with a high "G for Gamism" coefficient like Blades and Stonetop) with multiple, independent situations (including extended) going on and just smash-cut back-and-forth. Its just more cognitive load on the GM. Depending upon how well-slept you are (never in my case!) and generally how under duress you are outside of gaming, its either a "no big deal...lets do this" or a "I can keep my head above water well enough" scenario.
Unrelated, but necessary:
Needless to say, my position on the Story Now games feature or require a high level of illusionism and/or are shallow and/or are internally inconsistent
"critique" is utterly, 100 % baseless. That is 1000s of hours of experimentation & validation with dozens of fairly widely diverging games and replication with dozens and dozens of players (the overwhelming abundance of which were solid to great at the paradigm).
Anyone reading this or thinking about running a game that doesn't feature a metaplot or setting tourism or any type of top-down GM pressure to curate content/corral players via hooks/breadcrumbs (etc), its entirely doable. Its a real thing and you and your players are likely capable at it with a little practice or adjustment. Oh, and ITS NOT A WRITER'S ROOM. The roles are still GM and players and they each have specific roles (though different than normal). Roughly, the system and players create breadcrumbs for the GM to follow. The GM then frames scenes with those breadcrumbs as the constraints on stakes/goals/content and aggressively plays the PCs' opposition in-line with the parameters of system. Rinse-repeat. Story spins out of that deal.
You may not like it as much as games that heavily feature metaplot or setting tourism or other types of top-down GM pressure to curate content and corral players. You may like it more. But the two are different experiences. Try (and enjoy...or not) both.