Twitter has been around a little while now, perhaps it is even becoming mature, and certain patterns of usage are beginning to make themselves clear. Retweeting, the practice of forwarding on an interesting tweet to one's followers, has become standard and will soon be added by Twitter on their main site.
Another phenomenon that has entrenched itself is spam messages. This isn't surprising, every new communication medium quickly develops its own sort of junk mail. Luckily in the electronic world it is easier to take action against junk marketers than in the real world. Email spam filters have advanced tremendously, if you use a webmail service you should very rarely see spam in your inbox anymore. Twitter, unfortunately, is still in its golden age of spam with no filters instituted by the twitter servers and few third party services that help. So far I have noticed a few types of spam on twitter and for all of them there is a solution that each and every one of us can help with until Twitter gets its act together and institutes some system to deal with it.
- Spam Followers - Everyone on twitter is familiar with this type of spam, you get a message that you have been followed by @luciousVelvet, you look them up and see they have zero followers and either 1 following or thousands. The entire point of these accounts is to follow people and have a direct line to feeding through adds for their porn website or russian made cellphone offers. The best outcome for them is for you to have 'autofollow' on and then their messages show up in your feed.
Twitter solution 1 : Disable the ability to autofollow, its a dead feature.
User Solution 1: While you are trying to make out how @luciousVelvet could possibly have gotten into the position shown in her profile picture, look on the right side navigation bar, click "block", and click "okay". The effect of this will be that you have one less follower, but you also have one less fake follower who looks less legitimate now and from whom you will never see a message again, in any form.
Update: For even more fun, before you block them send out a message "@luciousVelvet #isaspammer" then some smart developer somewhere can develop a blacklist system for everyone with lots of #isaspammer tags on their account. Just don't misuse it otherwise we'll need the #isnotaspammer tag as well.
Twitter solution 2: Publish block lists or allow users to auto-block any user with more than x blocks by the wider twitter community. If you set x=1000 then it should be pretty clear that anyone blocked by 1000 people is a spammer and no one wants to see their messages.
- #Hashtopic Spam - I just discovered this type of spam recently while watching the fantastic explosion of geeky creativity that was/is the #songsincode twitter meme. Basically, once a keyword or #hashtag becomes popular enough bots will focus in on it and start sending commercials for amazing teeth whitening solutions with the tag. Spammers don't wait until a topic is in the top 10 trends, it happens much earlier, basically it seems if the topic is generating a message every few seconds its worthy to be spammed. This makes it hard to follow a topic.
Twitter solution 3: Add a "spam" or "offtopic" button somewhere on the twitter search result page so that while reading the results of a search, people can flag tweets that are blatantly off topic. Then allow people to select a setting that auto-filters tweets with a lot of "spam" flags on it. In addition, add a watch to the user generating those flags and use this as further observations in your advanced machine learning algorithm that decides when to disable twitter accounts. oh ya....
Twitter solution 0: Institute an advanced machine learning algorithm that mines various twitter data generated automatically and by users to decide when to block or freeze twitter accounts suspected to be spammers. If the freeze option is chosen then to unfreeze would require the twitter user to engage in a convincing email/skype exchange with some paid twitter staff to prove they are human.
User Solution 2: same as user solution 1, block spamming users. It would take too much time to block all the spammers you see, but we should all consider it our duty as Twitter citizens (Twitizens?) to block at least five people a day in the course of our normal usage. Then, if Twitter ever gets around to using or publishing block information, it will be useful for something. In the meantime, it will still be useful for you as you won't see any messages from those users again. Of course, they'll just create another account, but that's why we need Twitter to do something as well.
- @name Spam - This is a type of spam that I am not famous enough to have thought of (at least not yet...). I was reading Wil Wheaton's blog (yes,that Wil Wheaton, it's really well written) who's on twitter under @wilw and he describes what I will call @name spam or twitter spoofing (....twoofing?). This is where a spammer sends messages that have the user ids of a famous user in the message to get their attention or ours. Many people search for these famous users and will see the spam. There are a couple problems with this which expose some design flaws in Twitter itself (gasp!).First, Twitter recently changed @replies to @mentions on the main webpage so that instead of seeing messages that began with @yourusername, you now see messages where @yourusername is anywhere in the message. Users with lots of followers now see all these spam messages as part of their 'replies' and are overwhelmed, reducing their ability to respond, and destroying part of the community that Twitter is trying to foster.The second problem is that many of these spam tweets (twam?, spweets?, sorry, I'll stop now. ) actually misrepresent these famous users. They say
"Ok, I'll buy it! RT @ryanseacrest this waffle maker has changed my life! http://bit.ly/j3k2" when everyone knows Ryan Seacrest is actually more of a pancake man. This seems unfair.
Twitter Solution 4: All of the above solutions. Perhaps an additional, though controversial, help here would be to enforce RT message authenticity. This could be a bit stifling and expensive computationally, but in theory you could look at every message with "RT @bob ..." in it and check @bob's message for anything like what they've said. It would be easy to enforce exact RT's, but then people wouldn't be able to edit tweets to shorten them and make their own comment. I could imagine creating a metric that measures how similar the RTed message is to any original tweet from @bob and dissallow it if its score is below some threshold. But I don't think there is any ideal solution short of making RTs actual threaded links to messages so that users don't need to type in the original message at all, then the original message you are retweeting would show up below your own message on the Twitter page. Actually, that sounds pretty good, they could that, call that Twitter Solution 5, get to it Twitter developers!
User Solution 3: same as user solution 1, block'em!
All of these solutions may smack of censorship, but if implemented well, they could greatly improve the experience of Twitter while almost always filtering out only the users who abuse the system. In the end, spam will never go away as long as there are people who click on spam links and buy the merchandise. If no one ever bought things via spam, there wouldn't be any, that is certain.
Since we can't change human nature, I'll stick with suggesting my plan above to fix Twitter. Hopefully they'll listen, if they can hear it through all the spam noise. If not, then Twitter might become one of those dead technologies that was big for a while, but then became unusable and didn't adapt as quickly as google or facebook/friendfeed or bing/live/microsoft. Lets hope they figure it out.
But until then, we can all make our twitter experience a bit better by blocking people, start with just five a day. It'll make you feel good :)
PS: I actually looked up @luciousVelvet after I wrote this and it turns out its not taken as a user name! Stunning, maybe the spammers aren't as smart as I thought, or they're smarter.... Anyways, I predict it will be taken within a week.