Hey Google, don’t be evil!

Usenet news is the Internet’s original discussion forum (in fact, technically it predates the Internet, but that’s not important right now): groups organised into a hierarchies, with messages (articles) in them which have a similar format to email, but which are distributed using a different mechanism. You can read news using a news-reader, a piece of software which connects to a news server to retrieve articles. There are thousands of newsgroups out there, covering every topic you can imagine and some you’d probably rather not. A whole lot of the Internet’s subculture originated on Usenet.

A while ago, Google acquired a Usenet archive dating back for a couple of decades, and did their usual search magic with it. Unfortunately, they’ve since lost the plot. The latest revamp of Google Groups, which went live yesterday, has sacrificed the usability of the news archive to promote Google’s own discussion groups. These internal Google groups are presented on the same site, in a move which seems to be designed to compete with the Yahoo! Groups! service!. What’s wrong with the new site?<lj-cut text=”Bad Google, no biscuit.”>

In decreasing order of badness:

  • They’ve broken old links. There are countless web pages which cite Usenet threads by linking to Google’s archive. These are now all dead links. Old links to individual articles (which refer to the Message ID of the article) are redirected (hurrah!), but into the top of the thread which contains the article (doh!): good luck finding the article you thought you’d linked to in a thread containing hundreds of responses. Oh, and the new way of forming links to individual articles doesn’t use the unique Message ID which is already in the original article, but a unique ID of Google’s own devising. That ID is presumably subject to change the next time Google’s developers are bored.

  • Google are tracking clicks on URLs in messages (the URLs are re-written to go via Google). I regard that as somewhat sinister: why do they need to know which links you’ve clicked? They’re not doing that on their web search, presumably because they know what would happen if they tried it.

  • They’ve tried to obscure email addresses throughout the articles, replacing some of the local part (that’s the bit before the @ sign, RFC 2821 fans) with an ellipsis. They’ve ended up obscuring similar looking things, like mention of Message-IDs in the original messages. Obscuring email addresses in Usenet articles is pointless: spammers already suck up email addresses directly from Usenet servers.

  • In all these changes, they’ve still not fixed the problem of Google users replying to messages from years ago. Or, as far as I know, the other problems (like not honouring the Followup-to header) which mean Google’s posting interface isn’t a good Usenet citizen.

  • They’ve tried to be clever with fonts. The text of articles is shown in a proportional font (one where all the letters are not the same width). Some levels of quotation seem to be in monospace fonts (where they are). Mock HTML tags (like, say, <sarcasm>) seem to confuse it, for some reason, as do other variations in spacing. If you have a post which discusses HTML itself, Google gets confused and may render the whole thing as a huge link, or with random spacing and fonts.


On the plus side:

  • The new interface is pretty.

Google provides a valuable service by making the Usenet archive searchable for free. But it is also using this to drive traffic to their own discussion groups, and doubtless hopes to make money from ads, in a similar way to the Google web search and Gmail products. Unlike the web search or Gmail, though, Google has made a product by reproducing people’s copyrighted work in its entirely (except for those ellipses, naturally). In the past Google has got away with it because their product enhanced the experience of Usenet, which is, after all, full of people who want to be heard. Every change will produce some resistance from people who liked things how they were before, but in this case something has been lost, and pretty fluff does not make up for it.

4 thoughts on “Hey Google, don’t be evil!

  1. I agree with you on all but one point Paul. I do not like the new interface at all – to me its like someone let a five year old choose their colour scheme… Also it has this annoying effect of loading most of the page, then moments later a css style moves everything around – meaning you have to stop and wait for a few seconds to be sure its safe to read the page.

    Awful, absolutely awful.

    To be honest I don’t understand why they’ve done this. I doubt there is a great deal to be made from letting people create their own groups (presumably to get text advertising revenue), but whatever they get from that I doubt it will make up for the reduction in usage as a result of their insane design/functionality changes.

    (P.S. this is my first post from my new desktop install – Fedora Core 3 :0) )

  2. Subject: New interface is hard to use; results pages is crowded
    Well, i loathe the new interface of “Google Groups”, available on “groups-beta.google.com”; agree w/ robhu completely.

    Old interface is still available at groups.google.co.uk, and via some editing of /etc/hosts …

    216.239.37.147 groups.google.co.uk groups.google.com
    216.239.37.104 http://www.216.239.37.104 groups.216.239.37.104 images.216.239.37.104
    216.239.37.104 http://www.37.104 groups.37.104 images.37.104

    … see http://www103.pair.com/parv/comp/unix/cf/network/hosts.google for details/updates.

    – parv

    1. Subject: New interface is hard to use; results pages is crowded
      Dang! “216.239.37.147 groups.google.co.uk groups.google.com” does not work.

      – parv

  3. Subject: Click tracking
    > They’re not doing that on their web search, presumably because they
    > know what would happen if they tried it.

    Actually, they have done it for a long time with the web search – but only now and then. So for every 1000 result pages you get there are some with redirected URLs.

Leave a Reply

Your email address will not be published. Required fields are marked *