(Although the post may not be directly relevant to LW, I suspect I have an actual answer to the question.)
Speaking as a Wikipedia and Wikimedia volunteer of several years, I agree wholeheartedly with your concerns (though money is still lovely). The likelihood of disaster is not as high as it was, but many things are far too centralised.
I've been thinking about how one would fork Wikipedia, but have held off talking about it in e.g. my Wikimedia blog precisely because serious and acrimonious discussion of forking has been in the air, and it would come across as furthering the acrimony. Whereas I'm thinking more in terms of basic backup hygiene, against unfortunate meteors - physical or legal - hitting the Wikimedia Foundation.
So if you're feeling inspired to do something particularly useful, try this project I keep failing to get around to: spend your money on a spare PC with several terabytes of cheap disk, get a Wikipedia dump and see if you can actually use the software and configuration files provided by WMF to set up a working mirror copy of Wikipedia. And report back every step of the process to wikitech-l, so that the resilience of the encyclopedia can be improved - I can assure you people will be helpful and interested.
(The other two prongs of a fork are bringing a sustainable community with you, and having the legal structures to protect your project. Those are important things for someone to give serious consideration.)
I donated to Wikipedia before. However, this year, I'm not donating, and my rationalization is this: there need to be more Wikipedia clones on the internet, to best prevent the possibility that a single organization (Wikimedia) abuses its control of the biggest worldwide information source. Thus, instead of donating to Wikipedia, I would like to subsidize potential Wikipedia clones. Whether or not it has ads is of little concern to me.