Every team plays several sets of games on consecutive nights during a regular season. The cause is a scheduling problem with far more constraints than free choices.
The season has fixed endpoints
The regular season opens and closes on dates set well in advance, because the playoffs, the draft and the international calendar all sit behind it.
Inside that window a large number of games has to be placed. Divide the games by the available dates and the average gap between them is already close to two days.
Any date lost to another constraint has to be recovered somewhere, and the only way to recover it is to push two games closer together.
Arenas are shared buildings
Most teams do not have exclusive use of their venue. The building hosts concerts, hockey, college events and touring shows, and many of those bookings exist before the season schedule does.
A concert booked for a Friday removes that Friday from the team's list of possible home dates. Enough of those removals and the remaining dates start to cluster.
Clustering is what produces a back-to-back: two playable dates sitting next to each other with nothing usable on either side.
Travel geography compounds it
Road trips are routed to reduce total flying, which means a team visits several cities in one region before returning home.
Cities in the same region are close together, so the natural routing puts their games close together too. Short flights make consecutive nights physically possible, which is precisely why the scheduler uses them.
A back-to-back between distant cities is generally avoided. The ones that survive into the published schedule tend to be the ones where the flight is short enough to get a team into a hotel at a reasonable hour.
Removing them moves the cost elsewhere
Cutting back-to-backs requires either more calendar days or fewer games, and both have consequences well beyond the schedule.
Extending the season pushes the playoffs later and squeezes the offseason, which affects recovery, the draft and availability for other competitions.
Reducing games reduces the number of home dates each building sells, which changes revenue for the team and for everyone who works a matchday.
What the league can actually control
The schedule is optimised rather than solved. Software weighs travel distance, rest, broadcast windows and building availability, and no arrangement satisfies all of them at once.
The visible improvements have come from limiting the worst cases: fewer sets of four games in five nights, fewer long road trips that end on consecutive evenings.
Those changes shift difficulty around rather than removing it, which is why back-to-backs persist even as their distribution becomes fairer.