There seems to be some confusion about what LiveJournal.com's policy is on clients--whether they be automated bots, meme generators, journal backup tools, or anything else. This post will try to help clarify that policy, so you know what we expect from client authors, and what they can expect from us.
First off, read this page: http://www.livejournal.com/bots/. It explains much of our policy and links to most of the ways to get information out of LiveJournal.
In general, some big things that can get your client banned:
- Multithreading. Don't do it, please.
- Screen-scraping. You should never have to parse HTML.
- Downloading other people's journals.
As always, if you have a proposal to have a new interface for getting data that doesn't currently exist, please, post it here in lj_dev, get some discussion going, and we'll see about getting it implemented. We love seeing neat toys as much as other people, but our first and foremost priority is keeping the site alive.
If you have any questions, please comment here and ask, and we will try to answer them as best we can. Also, if you have general questions about writing clients, please see lj_clients, a great resource for client authors. Finally, if you have specific questions about whether something is allowable or not--post them here or email me, please!
Thank you for your support in helping us ensure the site stays fast and usable for all of our users!
 We are currently reviewing this policy to possibly allow journal owners to be able to specify that their journal can be downloaded by other users. However, until we reach a decision and such a system is created, please do not write clients that download other people's journals. Also, this doesn't apply to communities--those are available through the client interface like regular accounts if you are a member (maintainer to get comments) of the community.
Update:  Let me clarify downloading other people's journals. Tools like the LJ Markov Random Text Generator (pretty darn neat, by the way) that use someone's recent few entries to create data are okay--if you're using the RSS/ATOM stuff. That is a client interface, feel free to use it.
What's not okay is hitting someone's journal, scraping out comments, going back to the beginning of time, and creating local copies of someone else's journal. We would prefer people not to do that. Thanks!