Double or quadruple global channel capacities

What it says in the title. For those who don't know, global channels have a capacity of 2500 members. Now, while that might sound like much and more than enough for whatever you have in mind, keep in mind that it's the limit of total joined members, whether or not they are online or even still subscribed. Thus, a global can quickly be filled with players who play infrequently or no longer at all and lacking means to flush inactive players, it will can hit capacity even if only a handful of channel's members are still active.

I'm suggesting a straightforward capacity increase because it sounds like the simplest solution - the ability to prune inactive members would be nice, but that would involve inventing and implementing functionality to do so, while the current limit seems like it might have been determined due to hardware resources available at the time or was just an arbitrary number that didn't expect the game to grow quite so much over the years.

In any case, doing away with the low channel capacities would;

1. help alleviate the stress of constant channel migrations due to hitting the member cap
2. help welcome returning players, as as things stand at the moment, any active global channels they might have joined migrated in the meantime
3. facilitate cross-server communication, especially in terms of teaming, task forces, raids, contests and similar activities

So, standard code rant aside, can anyone think of downsides to increasing global channel capacities to five or even ten thousand members, or even high enough a number that "capacity" is a hypothetical concept entirely?

---

I'd honestly rather see the developers focus on exposing more chat management tools to channel moderators. Several of the large channels that have been "forced" to uproot, such as Triumph Watch or Victory Badges, would have been fine if moderators could have removed joined accounts that were no longer active.

---

Like I said, the ability to prune inactive accounts would be nice, but probably much more complicated to implement. This is more of a quick-and-dirty solution to the problem.

---