What you lose is the necessity to be critical of your own value assessments. Actually, you don't lose the necessity; the necessity is still there, but allowing > 101 votes lets you passively lose touch with it unknowingly.
Limiting to 101 votes is a way to discern the best from the good. If there are 500 delegates who I think would add value to the system, and I vote for all of them, I'm basically shoving the responsibility of discernment to everyone else unintelligently.(Slates deal with this shoving of responsibility intelligently, or attempt to at least.)
Let's say you vote for 200 delegates. Let's take a subset of 50, whom are both unelected and not very popular yet. When you look at the 101 delegates getting paid, you realize that you are actually voting for all 101 of them, but those 50 that are not popular yet actually give more value to the ecosystem than the current lower 50. You then think, "I think these lower 50 elected delegates are doing a good job and add value, but these up-and-coming delegates look much more promising. I'm going to drop my votes for the current lower 50 delegates to give a chance for the new ones to move in."
If you were forced to limit your votes to 101, then this diligence with discernment would have to happen automatically. After hitting 101 approvals, you would have to make value judgements on each new delegate you approve, comparing new approvals to old approvals, pruning out the weakest link. The reality is that only 101 of the delegates we like have the ability to actually be elected, and this needs to be a part of how we vote. What I'm suggesting is that we force this responsibility upon voters so to keep them organized with their value judgements, with the added effect of making a stitch in time to save nine. With the current system, our approval lists will get longer and longer, and this will become more and more of an issue, which will be harder and harder to fix.