Truth and Bogosity on the Internet

This might seem an odd topic for Macromedia DevNet—telling rumors from reality isn't really a technical subject.

But consider how much of what we learn these days comes from the net... how much of our planning and actions are based on what we learn there... and consider how the world is still coming to grips with even such basics of online life as email or online news. With text-messaging, phone blasts and other new communication mechanisms—among an ever-expanding population of connected people—we'll likely see rumors get worse before they get better.

The business angle? If you get sucked into a bad story then it can easily waste part of your workday. But if your workgroup gets sucked into a decision based on bad info, then you can lose weeks or months of development time. Anybody can lose time and reputation with a bogus story.

(For me, there's another business angle... I spend a lot of time reading and replying to threads like "Microsoft is buying Macromedia!" (what, again? ;-), or "Those Flash ads in the newspaper can see you in your underwear" or whatever... the fewer online rumors there are then the happier my life will be. But that's beside the point.... ;-)

We use the net to find true, useful things. Untrue and unuseful things get in the way. That's what the rest of this article tries to help fix.

Testing a story is a lot like debugging a project. The same troubleshooting skills used in technical work can help in debugging a rumor.

1. What did the speaker actually see? What's the source observation? What happened? Skip the theory and explanations for a moment... what did you actually see?

   In a computer we'd try to get beyond "it crashed" and get to the source observations about what applications were open, what operation was being done, what type of failure or dialogs were seen. In an online rumor the source info would be what the original speaker actually said, in their own full unedited words.

   Going back to the actual initial experience is the first key step in avoiding a wild goose chase. In 1993 we could not test facts presented by television or newspapers. In 2003 an active reader can usually beat TV to a story, and the game turns into observing which aspects the popular source highlights or ignores.

2. Can the initial observation be confirmed? Can you or others see it happen again? Can you help me to see it too?

   A single isolated report is hard to explain, because so many variables could have actually been the hidden trigger. A single eyewitness account can be explained in many ways. But if routine fails multiple times, or if multiple accounts describe the same events, then it becomes possible to start triangulating on the true hidden cause.

   For a software problem, a key step is in trying to make it happen on another system. A successful recipe for achieving the problem helps rule out cofactors from the configuration, and gives an engineer an achievable condition to test against. For a rumor, if the only people who have source info are those who want to believe in the rumor, then it's in a shakier position than if people who want to disbelieve it also agree on the source observations.

3. What's the hypothesis explaining the observation? Is it internally consistent? Does it fit the observable facts? Is there a way to disprove this hypothesis? Are alternate hypotheses available?

   It's easy enough to advance a reason explaining something, but just because it's possible to believe an explanation does not require that we must believe that explanation.

   One mark of a good, strong online rumor is that there is no way to disprove it. "Anyone who denies it is just part of the conspiracy" is a pretty unassailable belief system. Most religions are based on non-falsifiable hypotheses, and debates here can be interminable. "The timeout is due to network overload" can potentially be falsified by testing at an hour when loads are lighter... "this application cannot import PNG" can potentially be falsified by testing with a PNG sample file which ships with the application.

   An assertion which is structurally protected against disproof is not necessarily false, but finding a falsification test for the hypothesis is a strong lever towards either further work with that hypothesis, or pursuit of additional hypotheses to explain the observation.

It's the same stuff we use when programming or when troubleshooting a computer... just the same scientific method of going back to source observations trying to repeat and confirm them, tweaking small parts of the problem one at a time, drawing up successive hypotheses and testing these.

But why are online rumors so much harder to solve? Partly because the strongest rumors are juicy — they give some emotional charge to people.

That "Microsoft buying Macromedia" rumor comes up every six months because such an outcome would matter to people. Each time this rumor comes up the available facts are few, but the potential effects prompt people to talk about it and spread it further.

The attacks on New York City and Washington DC triggered whole collections of rumors, from the serious ("all planned by CIA and/or Jews") to the trivial ("tourist in rooftop photo didn't see the plane coming "). Dealing with such inhumanity is not easy, and entertaining a rumor lets us pick at the scab from different angles.

A robust rumor is also short and easily grasped, and the countering info is long and involved. "Search engines ignore SWF" is a great soundbite, and the actual story is much more involved. "Kidney-stealing gangs" is easier to convey than the evidence suggesting no such gangs exist.

Online rumors are also harder to solve because of the decentralized decision making. If you're troubleshooting a computer problem then the decisions are made by you or a small workgroup. On the net, though, each potential speaker is a new source for the rumor, and each needs to eventually reach their own decision on the veracity or bogosity of the story. In some cases there's positive incentive to be the source of news, and to quickly spread a story before lengthy investigation. The infectious nature of online rumors is a significant difference from the more straightforward technical problems we work with each day.

There's one more reason why some online rumors are so vibrant—not everyone necessarily believes the stories they spread. Some can find it advantageous to spread stories which they know they cannot prove. This has been true in the computer industry over the last ten years, and is increasingly true in political issues today. When a false story can achieve true real world goals, well, some people do regrettably fall to that temptation.

What to do about online rumors? Heck, if I knew, I'd be Supreme Overlord or something by now. But here are some tips for dealing with provocative stories online:

• Don't believe it just because it comes from a trusted source. Don't disbelieve it just becomes from an untrustworthy source either. It's just a datum or a hypothesis, and can bear examination on its own without prejudice.
• Use an online rumor repository. Isn't Snopes great? Some companies have even archived malicious rumors about them. If something is spreading, then it has already left a trail on the web, and others have likely already created resources countering the fiction. (Having a list of common logical fallacies at the handy is helpful for recognizing structural flaws in stories, too.)
• Use alternate news sources to find additional observations or hypotheses. Google News is great for this, particularly for political news, because you can quickly pull up articles written from divergent viewpoints, increasing the chances you'll find facts which don't fit the original hypothesis. A newspaper you've never read before is only a click away. News sources you despise can be particularly valuable in finding holes or vulnerabilities in your own understanding.
• For breaking technical news, bloggers will tear apart a story from multiple angles. David Sifry's Technorati service can search on an URL or text term, and commentary is usually listed in reverse-chronological order for finding root citations. While 90% of everything is still crud, the remaining 10% can be a very efficient way to boost your understanding of an issue.
• Instead of trying to persuade someone they're wrong, have them try to persuade you they're right. "You haven't made a strong case to me yet... you have not yet successfully persuaded me" efficiently shifts the costs to where they actually belong. (Watch out for people who try to argue by flooding you with info rather than getting to the core of the issue... "If someone can't get to the point then it's often because they don't have one" is a handy rule-of-thumb online.)
• Make fun of people who spread such rumors. Seriously, show some tough love here. Someone who comes in weeks after a popular rumor has been widely debunked with a "Hey I just heard..." hasn't been paying attention, and has no claim on your own. Assuming you see no evil intent in their actions, then poking gentle fun at them may be the kindest way to help them avoid similar problems in the future.

We who specialize in creating the network have a responsibility to help the rest of those coming online in using the net efficiently and to the greater good. I'd like to hear additional thoughts you have on this subject, and will open up an item on my weblog for comments. Thanks!

 

Submit feedback on our tutorials, articles, and sample applications.