Hari ?page= is necessary for search engines to browse down a discussion. If we set every page with canonical URL to the first post, the search engines will ignore the next page links as duplicates.
If you open a discussion on the first post and check the source, you'll see the canonical meta is the link to the first post.
With ?page=, every group of 20 posts has its own canonical URL and can be indexed separately by search engines.
Ideally we should use the same /<post number> syntax that we use for javascript users, but it's a bit complicated to implement for search engines without ending up with as many pages indexed as there are posts. (because anytime a post is deleted, the number for the start of each page would shift, which would cause completely new URLs for each subsequent page if we only allow search engines to index numbers that start a batch of 20 posts)
