Spell checking is already "decentralized" and has been for the last three decades or so, in the form of a local dictionary file on each machine you want to run spellcheck on.

I wonder what would be a good storage structure for the dictionary? Oh.. a dictionary.

Probably a trie, although tries are sometimes implemented with dictionaries.

I don't actually think it should be, but I can make a compelling case.

While it's possible to have a dictionary of all of the words of a language, it will always be missing the proper nouns. It will also be missing the data needed to take my very poor spelling and figure out what I'm trying to say.

Combining these two problems, I've found that oftentimes my local spell checker can't figure out what I'm trying to say, but Google can figure it out no problem.

I don't mind having to go to Google once in a while to do this manually, but that can is the case for a distributed spell check service.

Doesn't Google use a large corpus of documents along with machine learning/statistical analysis to do spelling and grammar checking? The same way they do language translation.

You are confusing index building with index lookup.

Yup - this is my fundamental problem. HTTP or HTTPS isn't the issue - sending everything I open in my editor to a third party is the problem.

Hadoop cluster?