My frustrations with the cluster go back at least 20 years. Sadly not a lot has changed in that time, and DX Summit is one of the worst offenders. The issue is a balance between a spot flood vs improved accuracy. When I first started out I just picked the physically closest cluster (trying to avoid latency problems), and put in a few rudimentary command line filters from a complex set of options. This proved unwieldy after a while, and soon Lee VE7CC created his CC User program, which allows painless GUI-based changes. What's more, using the latest version of the software, you can connect multiple clusters and it aggregates the feeds, which is neat.
Less well known that his CC User software is Lee's actual cluster software called CC Cluster, which is a competing product to AR Cluster and Linux-based DX Spider clusters. Why is this important? CC Clusters allow even more granularity of filtering as well as turning spots on and off on a band-by-band basis based on the time of day, based on your inputted location details. Since CC Clusters are paid software there aren't many people running it, but Lee has been running his own cluster on it for years, and by default that's the cluster I use even though it's thousands of miles away from me. It's just that reliable! Almost no downtime, and it automatically rolls over to another CC Cluster when it is down.
So I have typically three clusters connected this way, and in turn that feed is available to DX Lab's Spot Collector and N1MM. Spot Collector itself is good, but I prefer feeding it from CC User. Of course spots are still very much a case of GIGO (garbage in, garbage out), but this process has been helpful.
I would strongly back the creation of a "Cluster 2.0" network with fully authenticated user accounts and sysadmins who are not afraid to block abusers.