Online surfing suffers from broken links. They pollute the web. The best way to prevent these annoying “thorns” is to reduce making them. We have to cope with the recklessness behind manufacturing them. Taking the problem personally and seriously means finding ways to bypass the need for links, which work perfectly until they break for some reason. For instance: Instead of linking to Wikipedia in order to mention a source of a quote it might be enough to refer to the heading of the entry, which is usually a unique keyword.
In order to fix broken links I have to click on every single link on my large web site. When the link doesn't work or works too slowly I ask myself if I can delete it. Most of the times the answer is positive, and I ask myself why did I make the links from the start.
It might seem too tedious to fix all these broken links one by one without using any software, but I can tell you that after a while I found patterns that gradually helped me find shortcuts.
I don't believe in solutions from “above” (that Google would encourage us, up loaders, to use less links and rank higher). I even don't mind if nobody else but me genuinely thinks that it's worth the time to fix broken links. I'm just convinced that's what I have to do without delay, and I share it with those who read this.
Tip 1: remember that a link is only a highlighted text. If you have a lot of broken links on one of your pages you may consider canceling the highlights (copy the document to MS Word with the Special Paste command unformatted text). Without highlights a link becomes regular text and no one can break it.
Tip 2: consider writing textual (not highlighted) links. Readers who will be really interested in following these links will have to make the (copy and paste) effort.
Tip 3: avoid links to PDF files, which download too slowly.
Tip 4: avoid dynamic links.