In its infancy Twitter created a widget for its website called trending which showed the 10 most common hashtags used at the time this was #useful but it was also #limited. For one, it only showed hashtags and didn't cover things which had not been tagged. The other thing that made it limited in its effectiveness was the fact that it was purely volume based. If enough people tweeted the same hashtag it would trend. With a handful of accounts tweeting the same thing in quick succession you could make something trend inorganically.
Over the years Twitter has refined how hashtags and trends work. The trending topics on Twitter no longer need to be hashtags, natural language is recognised. Hashtags still exist as an easy way to make tweets searchable in a feed. So too do cash tags or stock tags, like $AAPL which act like hashtags but are supposed to be used with the stock ticker of the company you want to tag - the site doesn't prevent you from using nonsense.
Beyond the natural language analysis, Twitter also expanded the scope of its trending algorithm to analyse the tweets being made. For something to trend it needs more accounts to tweet about it rather than a small number. You also need multiple plain tweets rather than one tweet simply being retweeted many times. Volume alone no longer makes a topic trend.
The trending algorithm is not without its limits however. It is still misused by some companies and organisations. Large corporations with many accounts for regional variations for example can engineer trending topics by coordinating the tweets of those accounts. One example of this can be seen in the UK with Heart Radio who regularly use this approach to make topics of their choosing trend. This behaviour however is actually against the rules of twitter, and if Twitter realises your actions as an individual you are normally suspended from the platform for doing it. With corporations like Heart if Twitter chooses they can sue for loss of revenue as this circumvents Twitter's sponsored trends feature and amounts to unsolicited mass advertisement on the platform - a.k.a. Spam.
Beyond these issues however there are instances where news items on popular sites are shared via twitter, where thousands of users click the tweet button on the website and post the tweet without editing it. This commonly results in all or part of the headline of the article trending. Personally I would prevent this from happening and amend the algorithm to only include tweets which have been authored by the user rather than using pre-written tweets and links like these. It's often the case that you see something trending, click the topic, and you see thousands of tweets all tweeting the same news story from multiple sources - something which I saw someone on twitter refer to as "churnalism" which I thought was apt and quite clever. I'd not heard of the term before but it can be seen quite clearly in these moments.
The problem I have with the way the algorithm works at the moment, is that it's hard, nigh on impossible to filter out all those tweets that are just the headline and the link, to get through to the tweets people have actually written about the content, or their reactions. In other words the third party content drowns out the social content. The tweets you see in those feeds are all pre-written by the third party media outlets, and very few, if any, tweets are actually what people on twitter have written in response - if you can manage to see them at all.
No comments:
Post a Comment
All comments are moderated before they are published. If you want your comment to remain private please state that clearly.