The Collision Effect
It's often better to get out of the way and let two stacks collide when ICM is extreme, says Dara O'Kearney.

Often, satellites end when one player literally blinds out. They are folding everything in the hopes of outlasting another player, and they get down to 1bb, before busting when they post their blinds.
This is actually sometimes the correct strategy.
I often get asked, therefore, if it's a good idea to limp to bust the micro stack when the blinds force them all-in?
The answer is, not as often as you'd think.
Why limping against micro stacks backfires

The solvers tend to avoid doing this without very strong hands, think 77+ AQ+. Or much wider if you are the SB just having to complete.
First of all, this is a bad idea if you are somewhat vulnerable yourself. You should only have chips to spare.
Secondly, if there is any chance one of the other players is going to shove over the top of you, whether it's because they have a hand or simply they are spite shoving with a chip lead, you want to avoid it.
If you limp, it might also put off another player from VPIPing with a stronger hand, as they will be worried about you.
But mainly it is simply because it's a bit of a disaster if the micro stacked player triples or quadruples up. It is game on for them and will extend the bubble for a long time.
This is a concept called the Collision Effect, where folding is better if it increases the chances of the other players clashing.
Let somebody else do the dirty work and limit the micro stack's upside.
You can get Dara's free book on effective poker study by joining his mailing list.