darobin
commented
9 years ago I certainly won't argue against the value of links. But as you say, there is a whole craft to the "how" of links, and that is definitely part of the issue here.
Groups are well-linked (notably from TR) and as a result are more findable through search. They're hardly perfect though: if I want to find the master list, I can either hit the W3C homepage where it's buried below the fold on a busy page, search, or remember that W3C lists groups under "activities" and use my browser's history (if I search that for groups it finds a lot of other things but not that page). This surfaces an internal jargon problem: "activities" is something no one outside the organisation knows about, and we sort groups base on their types whereas people are likely searching by content.
IRC logs are at the other end of the spectrum. If you know the date and the URL structure and the channel name then you stand a chance — otherwise you're pretty much on your own unless a kind soul has used some form of meaningfully labelled link to the log.
Even worse than IRC logs are charters. At least if you don't find an IRC log you're not going to get wrong information. Charters are given unique URLs per version, which means you could find the wrong one. It is not rare for search engines to find the wrong version first, or for that matter for group pages to link to the older version. Since we let groups lapse (sometimes for years) an expiration date in the past gives you no useful information (neither does one in the future as we sometimes close or recharter groups early). Case in point (the very first one I tried): http://www.w3.org/2013/07/webappsec-charter.html. This is expired, but is it the ongoing one? It even tells me to look for a revised charter but there isn't a link. I have no idea how to find out. This isn't an anomaly, it's a recurring issue. And charters sort of matter for IPR…
Automating links to these types of resource is precisely the reason behind collating the information relating to a group at a common location and unifying the manner in which group pages are produced. It's a great job for automation, and a terrible job for humans who naturally screw up on tedious repetitive tasks. So I think that on the front of increasing link density and automation, we are in violent agreement (and the notion is supported by the report).
Linking does not solve all issues however. In terms of information management (and also to help people find things naturally, through their address bars for instance) it would certainly help to modularise our content rather than have one single sprawling site.
Many times in the document the author complains that he can't find things. W3C groups, IRC logs, and so on. Just complaining you can't find things is not enough -- you have to say how you tried (following links, Googling, guessing URLs?)
Take working groups. "Impossible to find". There is a list here http://www.w3.org/Consortium/activities. What is wrong with that? Whenever you (anyone) says you could not find something on the site, please say where you were looking!
Links are a strong solution. If you make links, then Google will find things too. But when you make a link there is a huge art in understanding
this means taking your head and putting it into the point of view of the audience, for each audience. This is the art of communication in a hypertext world.
In a lot of cases links can be made automatically in both directions. Tools should do this when they process stuff. In some cases, people will hand-link it one way and and a machine can find that link and add a matching back link. E.g. search for all pads which link to the group and make a list of them in the group.
So IRC logs for a group or a channel should be linked from a master list for that group or channel. Recent IRC logs, pads, messages , chats, discussions should be in a notification stream on the home page. Each log can link to the next and previous logs and to the main list. That can be done automatically.
Often it is easiest to link in one direction than the other, but both are valuable.
Other techniques:
In general these links should have some semantics so they can be stuck in a data store and queried.