Generally, this decision by Ancestry is not a problem, at least for most purposes. (See IBD, iBS, or IBC?.) Matches between 8-cM to 10-cM are suggested to be Identical by Descent (IBD) about half the time. Matches below 8 cM are more likely Identical by Chance (IBC).
However, sometimes those small 7-cM matches on Ancestry can and do mean something, especially when many other people share the same common ancestors with that person. Sometimes when I check with one of my cousins, I find that they share significantly more DNA with that same person, more or less confirming that the match is IBD. Since I have been exploring Cluster Analysis programs, I have come to realize that the small matches can be more meaningful, when shared with several other people who are cousins. I feel that such matches may eventually lead me to find distant ancestors who are presently unknown to me. In that sense, the removal of small DNA matches will certainly make some of those connections harder to establish.
I have two takeaways from this:
1) Matches with notes, group tags, or messages will not be removed (although their ThruLines may disappear.) So, I am adding comments to any matches whom I want to keep within my match list., which should keep them available to me.
2) This decision by Ancestry underscores the importance of Gedmatch.com for DNA comparisons. Gedmatch also seems to be more accurate than Ancestry. Gedmatch often shows larger matches than Ancestry, although occasionally much less. If everyone uploaded their DNA to Gedmatch, the impact of Ancestry's decision would be minimized.
When dealing with statistics, even rare events can and do happen every day. In the case of small DNA matches, the trick is determining which of those rare events is meaningful. That is certainly a neat trick, but not always possible to perform.