Any of my frequent readers are likely to know that I am a slashdotter. Slashdot, for all its faults, is still by far the finest geek-oriented news aggregator and forum around, at least in English. Slashdotters have long had a battle with Slashdot over story submissions. Duplicate stories abound, and so many press-releases-as-stories have shown up over time that a term has been coined for it: "slashvertisements". The result was that Slashdot introduced the Firehose, in which users (paying or not) can vote for or against story submissions and mark them with descriptive tags.
When I started reading the firehose my first reaction was one of revulsion. Are people really this stupid? Absolutely tons of the submissions have no link, and of those that actually have them, most of the links are absolutely atrocious. This and several other typical failures of story submissions prompted me to write a Slashdot journal entry entitled Slashdotters need Help with Story Submissions in which I railed against poor HTML link etiquette and other issues. Obviously I have some extra time on my hands.
But I want to talk more about link etiquette to the internet at large, because it is a serious problem with the web. At this point many of you are snickering, but those of us who actually try to use the web have discovered that bad links make it harder to find information. They sabotage the efforts of search engines, which use the link content to classify the type of link, which in turn is used to classify the type of page that the link connects to.
Let's talk about some common ways to provide a URL to a user and what is right or wrong with them.
Non-Links
The first most common way that links are mispresented is by providing a URL without a link. This is done by simply typing the URL into the text: http://example.com. There are a number of problems with this. First of all, search engines cannot determine your intent, so they must either ignore non-link URLs, or make a potentially incorrect guess as to what you meant. Second, if the URL is given at the end of the paragraph the user must determine whether your ending punctuation is part of the URL or not.
Links whose content is the same as their HREF property
You might think that the appropriate solution to the above is to simply make a link whose contents and href attribute are the same: <a href="http://example.com">http://example.com</a> will produce a link that looks like this: http://example.com. It's valid HTML, and it is a link, but it is an example of a bad link.
Sometimes you DO want to make a link like this, such as when the URL appears in the text for a good reason. This might be a result of being part of a press release, or a page of contacts. The URL should be legible and users should not have to mouse over it to see it. But in every other case this is not appropriate.
Links whose content is "click here" or "here"
Another example of a bad link is exemplified in the following example: "You can find the information on examples at example.com. To access example.com, click here." This is a ridiculous way to link to some information. For one, the user may not be clicking; it could be a tap with a stylus, or a press with a finger. They might be using a text browser with no mouse, and they might be tabbing or arrowing down to the link and then pressing the right arrow. Device independence is very much the point of HTML, and so you should not make assumptions for your users.
But simply changing it to say "here" instead of "click here" misses the point entirely as well, because that is not the only reason not to make a link like that. The content of a link is meant to be descriptive of the content or type of content that you will find when you follow it. This is important for search engines (which use both the context and content of links to characterize the pages to which the links are made) but is also important in a number of other cases, especially now that the vast majority of webpages consist of managed content. A content management system is able to generate additional content on the fly based on various sources of information. Slashdot in particular places a block of links garnered from the story submission into the sidebar so that you can easily find them. When the links have descriptive content, then this block provides useful links. Otherwise it becomes a confusing hash of links titled "article" and "here".
The proper link
So what makes a good link? Context and content. consider the example from above: "You can find the information at example.com. To access example.com, click here." It's so very easy to make this particular bit of text make sense, and you can shave an entire sentence away and come away with a more accurate message. It works like this: "You can find the information on examples at example.com." In this case, you might even make the text "example.com" a link to the root of the site, as the link that says "the information on examples" goes to a specific page. But ideally, you would have done that someplace above this text, introducing the information.
Links whose content is the name of the site and not the page
This comes after the proper link because I need an example of a correctly-made link to screw up and pick on. A lot of people, given the above URL and a need to link to it for some examples, will make the following mistake: "You can find the information on examples at example.com." This is one of the most common link mistakes I've seen. It implies that the link to examples.html is a link to the front or main page of example.com, which is of course incorrect.
Remember, link text should always be descriptive. No links to "here", "click here", or any of that nonsense.