Cancer, in case you haven’t heard of it, is a horrid potential byproduct of the natural process of cell division. Things go wrong, a process goes out of whack, and the next thing you know you’re doing all sorts of insane, crazy shit to fix it. And make no mistake; most cancer therapies are insane, crazy shit. The results are not always great, the earlier you find it the better (because then the madness is in one place) and in the meantime things are generally horrible.
The real bad part is, you can get cancer of the nearly everything. Unlike diabetes or migraines or breaking a bone, the problem can, quickly or slowly, spread elsewhere and make cancer happen there. It is pernicious and deadly.
Kind of a bummer, I know. Let me talk, then, about online cancer.
A while ago, I heard Tom Jennings say a brilliant thing. This is not hard to do, if you’re listening to Tom Jennings. But it struck me how many levels of brilliant the statement was, over time, considering it from many angles. It was at a Vintage Computer Festival talk, the one where I beta-premiered the nearly-done BBS Documentary episodes. The organizer, Sellam, had decided it would be fun to have the Fidonet guy also speak at the same event, and then have the Fidonet guy sit through an episode about himself. I do believe that was the scariest part of my entire production!
Anyway, Tom gave a talk about the considerations of putting together Fidonet. And he mentioned some error checking, and he veered off into a rant about error checking. And what he said was:
“Error handling is 90 percent of the work. Most of your time is spent trapping for everything that could go wrong.”
This may on its face sound either simplistic, wrong, or exaggerating, but once you consider it thoughtfully, it blossoms into brilliance. Rephrased another way, it could be considered that engineering is primarily the preparation for unwanted events both external and internal, not just the successful completion of a series of wanted events. The “holy shit it works” phase is rewarding and heartwarming and all the rest, but under controlled situations with little variance in user pool and approach, stuff can be made to work very quickly.
A buddy of mine worked on a lot of software projects, a well known instant-messaging environment. He told me how they’d check in a new revised version of the server and they’d know if it was working within 90 seconds; that’s how long it would take for the onslaught of user interaction to slam it to pieces. It’s not the “hey, people can message each other” part that kills you, it’s the “everybody can do anything and that includes stuff that wasn’t accounted for” part.
A lot of time has been spent on making sure stuff works, and can handle the onslaught of stuff that specifically makes a website crash. It’s rather hard now to put in some magic set of control characters or paste in HTML or pile on the long-length words and make a site go down. When you’re faced with an input window of text that you can fill out, chances are you won’t cause the thing to die. The problem is exacerbated by the use of libraries, where the libraries are created by an entire other group of people outside of the ones making applications that use those libraries. Applying my rule of the two-way street, this is fantastic in putting in improvements — everybody’s stuff gets better! But it’s disastrous when there’s a security bug — stuff you didn’t even know used that library is now able to be dressed in a gimp outfit!
But all of this is well-tread ground. If you base the success or failure of your project/application/website on its ability to stay stable and online and handle whack-ass input requests and poorly-formatted queries and the rest, then you will have success relatively easily, and this “start to success” time has decreased as libraries are strengthened and people depend on the strength of years-and-years of quality control. There’s an entire industry dedicated to software-level quality control and a lot of effort put into it.
So where’s this online cancer I’m talking about?
My web browsing puts me into a lot of forums, in a lot of websites, in a lot of places. Sometimes these forums are fully-formed groups of people with a specific theme. Other times, it’s just the commentary at the bottom of an article or weblog posting, or under a Youtube video. They’re all, in some way, forums, just at different levels of complexity and thought given into the process of people being able to communicate with each other.
Obviously, I’m looking at these places with a more historically-aware eye than most. I passed my quarter century of using “message bases” last year. In fact, let’s pull one from that time:
http://www.textfiles.com/bbs/PRINTOUTS/198408shpolitics.txt
This one is interesting, because it’s about politics, always a contentious issue. Yet, generally, discourse is outward-focused (Reagan, Mondale, Heavy Metal) instead of inward focused (users, the political structure/system, the quality of posts). The reasons for this are pretty clear: very small group, and very long turnaround time. Remember, only one or two people can use the Safehouse at once, and so the first post is on August 3, 1984 and the last post is September 10, 1984. 38 days later. With 26 total posts, that’s an average of less than a posting a day. 24 hours between posts.
Now, compare that to a thread on Fark.com. Here’s a good one:
Ex-Marine kills bear with log.
This thread, meanwhile, started at 9:04am on June 21, 2007. Within 12 hours it has 187 messages. The majority of those 187 messages are posted in the first 3 hours.
So we go from a posting rate of once every 1,440 minutes to once every 45 seconds. In the first example, it’s about “politics”. In the second, it is “An ex-marine fought off a bear with a log.” and a link to a news story.
The conversation, and bearing in mind this is one of sixty new stories/topic threads started on Fark the same day (two an hour, every hour, the entire day), starts off with jokes and commentary on the story. Then it divests into discussions of camping and marine egos. Then it turns, discussing the motivations/choices of the marine in the story, and the wisdom of the other posters. Bear in mind, too that Fark implements moderators who remove particularly offensive or attacking letters, so this represents, if not the cream of the crop, the best of the remaining.
Keeping up with all postings on all threads of Fark would be a full time job. There’s another online community-thread site called Metafilter. They had 22 story postings for the “main” page (there are sub pages), and over a thousand comments between them. Another full-time job.
Fark and Metafilter are both working very hard to control what I’m talking about, the online cancer where the pure mass of postings into a forum from so many disparate folks inevitably leads not just to bickering and misunderstandings but screaming, blind hatred within an hour. Fark and Metafilter have subscription levels, inexpensive, but functioning as a barrier to entry (in the case of Fark, to see topics before they’re posted, and in the case of Metafilter, to be able to post). Metafilter, in fact, has gone years ahead of a lot of websites and has moderation of comments, history browsing of users, and a very, very strong ruleset about quality of topic threads, duplicates, and so on.
I don’t think a lot of people are recognizing this cancer for what it is, this rapid, rapid posting of messages where any out-of-lockstep post can send the entire conversation flow cascading down into bitter side-taking and attacks. Mention a political party. Mention a hot-button topic from recent news. Mention a place, a job, a race, and watch as everything turns on itself and makes a horrifying feedback noise. As much as actual cancer comes from too-quick division of cells, I feel like this online cancer comes from uncontrolled, unchecked cascades of off-topic messages with no real moderation/quality control capable by the site or the users themselves. Sure, a place will send their guy in to delete an article with someone’s social security number or which informs another user that they are going to be killed in their sleep; but that’s the level of baseline legal protection. It’s what you do because you don’t want the site sued. But what about the quality of the conversation, the working together to end up with a conversation that makes everyone who plays a part in it better for contributing?
Choose a youtube video. Choose any popular Youtube video and you will watch the cancer eat out the bottom of the page with people turning on each other, insulting each other, dragging topic into places that make no sense and have no bearing. And there you go, someone might have spent weeks working on a great video and below them lies absolute garbage. Sure, they can turn off comments 100%. Is that really a solution? No.
Will the conversation and the engineering really turn its priority from ensuring that there’s no buffer overflows in input windows to coming up with a lightweight “personality” that you can have from site to site? Or devising improvements to moderation, or determining solutions to this online cancer? Cute little conferences are being held around the world lightly touching on this, but the idea, which should be as fundamental a goal and question as getting the webserver machine running, is an afterthought, a hindbrain twitch, a quick sleeve-polishing of the glasses before throwing it out to the wide world.
It shouldn’t be. The cancer is going to grow.
4 Comments