Last month, #AmberTurd secured one of the top spots on the Twitter trends list, and one of the most used phrases in the most-watched courtroom of the year.
Yep, the Johnny Depp vs. Amber Heard case.
So, we’re not digging into the case or its justice implications or the like.
But the hashtags.
The gravity, weight and relevance this trial gave to hashtags seemed unprecedented.
We’re not lawyers here at RedMark, and don’t spend much time in courtrooms. We DO, however, use metrics as a way to indicate a possible outcome.
So in the same way the lawyers wanted to link these hashtags numbers to indicate negative public sentiment, a business might look to link certain hashtags to interest in their brand, product or service.
And while hashtag analysis can be a helpful element in some metric reports, the Johnny Depp trial highlighted three reasons why it can also be a slippery slope.
Twitter Bots are getting a lot of attention lately from Elon Musk’s perspective, but they were a topic far before that story. Five years ago a Pew Research Center study found that 2/3 of tweeted links were shared by “suspected bots” and about the same amount specifically for links about news and current events. That’s a high percentage even if some of that “suspicion” is false. Given this, it is shaky ground to assume that tweets, and thus the hashtags within the tweets, are all being posted in real time by humans, and therefore a reflection of the market sentiment.
In the trial, the social media experts claimed that they had filtered searches for specific hashtags and then showed graphs about the number of tweets that contained that hashtag. Not much was made of the fact that they didn’t know the actual context of the tweet. Assuming context from a simple hashtag can be tempting, but dangerous. If users are tweeting your business name as a hashtag, for example, it can be equally possible that the overall context of the tweet is positive or negative.
Turns out, we’re just meaner on Twitter. When we look at metrics, we have to know what we are baselining against. You couldn’t compare all open rates for an organizations emails, for example, against the open rates of one campaign giving away a free car. It’s just not apples to apples. Similarly, what we say and how we express what we feel on social media is not the same as how we speak and express sentiment in the real world. John Talin, USC’s director of the Anneberg Innovation Lab, has researched and analyzed attitudes behind Twitter speech. The lab’s “Twitter Sentiment Analysis” found a disconcerting disparity and increase in hostility, vile language and negative attitudes on Twitter. Certainly, we’ve all heard of “keyboard warriors” who “hid behind their computers” to spew vitriol that they generally wouldn’t have the guts to put their face behind. If this is a truer reflection of their heart condition is not necessarily the point to be dissected. The baseline for cruelty, for slander, for negative hashtags on Twitter, however, should certainly be addressed before placing evidential weight on a hashtag campaign.
What did you think of the HASHTAG analysis from the Depp/Heard trial? Were you #skepticalofhashtagevidence too?