Observations from a noob.
I'm a noob at the thons. I decided to try this one in full spirit, keeping an open mind and trying to reserve judgement - but concerned about the quality of the entries - everything screams quantity over quality. I intentionally did not use any bots or browser-automating code (maybe next time). Just pure editing profiles mostly from my list of things I already work on. I worked a lot over three days, including an all-nighter and a nearly-so, and got north of 100. I think I worked fairly efficiently, but it was necessary to do some additional research every now and then 'cause the data just "smelled". The research in particular kept me from reaching a higher number.
From my corner, there is benefit. I was motivated to put a stake in the ground on many things on my todo list, and also I'm seeing some new connections (which is touted as the point of the exercise).
The amount of hours I put in is not sustainable. That's true of probably all thons.
I think my concern about the quality of the results is warranted. I have seen numerous serious errors in some of the profiles I've checked out from some of the highlighted winners of this event. A few things I see after the fact: profiles with <ref> tags inside the references section: fix by moving EVERYTHING into research notes; profiles citing only an unsourced findagrave (which has significant errors); generated TABLES (boo) in the bio section which amount to beautified copy-paste of census records and an intimidating-to-edit bio section; data dump disguised as a biography; unreadable references with no punctuation, just a bunch of words strewn together; mis-attributed family; references with too much data from the original source; conflated data; more. It'll take a lot of work to dig out from some of this. More likely, many of these profiles will sit exactly as they are for a long time.
Subsequent to the event, I have learned and tried some automation tools, and I can see some advantages. However, I think it can be like giving a chainsaw to a five-year-old. A lot of stuff is probably going to get cut down, very efficiently. Hopefully the right stuff, but not always.
The options in the wikitree sourcer can be set in such a way as being suitable, IMO, for this kind of activity; however, many people are using the tool in a way that sets up problems, including excessive formatting of what's essentially an automated entry (which is just another version of the gedcom import issue, and if the data can be found in the source, why copy-paste it to the profile?). The wikitree style guide page that says plain text is almost always better has only been accessed about 5k times, so probably most people don't know it exists. Nobody likes hardfast rules, but a "guide" that is going to be ignored might as well be erased. If someone's going to enter 1k profiles in 3 days, the resulting profile should probably be as simple and approachable as possible for all - not customized beyond the recommendations - especially when there's already instruction from this thon on how to remove profiles from your watchlist and trusted list.
On the fence about whether to participate in a future connect a thon. I'll probably follow the results a while and see what happens with some profiles of interest.