Last night I published over on my personal blog my thoughts on how Twitter as a public company can turn around their fortunes. I laid out four main areas I see where they need to work – messaging, user identity, making decisive decisions, and combatting abuse – but I feel like one of my strongest points got lost.
Twitter needs a Panda update.
If you want a full background on this algorithm from Google, you can find it here, but quickly Panda was Google’s answer, starting in 2011, to their issue of low quality content farms (think eHow, Squidoo, etc) that were gaming the algorithm hard but not providing a good experience to users. Starting in 2011 and over 20 times since then (and now it’s baked into the core algorithm), Google has run this algorithm to continue to deal with sites that present a low-quality result to the Google user.
Of course, I am not naive enough to think that they do this out of some drivenness to do what is good for the world. They do/did it because they need to keep their market share to be able to sell ads against it, which then allows them to invest in changing the future with things like cars. If your users aren’t happy, they won’t click ads.
Facebook did the same thing back in 2013 to combat virality mills like Upworthy and more recently around Facebook Pages. Sites like Upworthy and others were gaming the algorithms hard to drive insane amounts of traffic back to their sites to monetize through ads (same thing sites were doing with Google pre-Panda), and that was a bad experience for Facebook users. Once again, make your users unhappy and you can’t sell as many ads.
It’s time for Twitter to do the same thing.
Twitter’s Difference
Twitter has a huge spam and abuse problem. Spam because they have allowed so many different apps to post anything they want to the platform, so many Twitter feeds are just a collection of links. I know – my company Twitter account is currently this way. And it’s worked – I’ve gained quite a few followers (and engaged ones) by doing it. And I love companies like Buffer who allow this (I’m a paying customer), but there is a downside to automation like this.
So Twitter could implement some changes (and I think they are already working on this with features like “Things You Missed” that you see within their native app at least) that would reduce the reach of updates like this. It really wouldn’t be that hard to do with some basic machine learning and some work on their core algorithm (yes, they have one).
Automation is the least of their worries though.
Their bigger issue is abuse and harassment. Just yesterday we saw PharmaBro Martin Shkreli get kicked off Twitter for harassing a Teen Vogue reporter. That’s one of the few examples we have around people actually being removed from Twitter (they also removed a bunch of Neo-Nazi accounts), but it’s sad that it took the level of Shkreli’s harassment for him to actually be removed.
Those levels of harassment are fewer and further between than the abuse that a lot of people, especially women, put up with daily on the platform. Just check out this (sorry, somewhat NSFW) Twitter search for “I will rape you”.
Ghastly. Unconscionable. Unacceptable in a civilized society.
And yet ALL of these accounts are still active! These accounts, saying that they will rape someone, are all still live on the platform.
Twitter’s problem is admittedly a bit tougher to solve than Facebook or Google’s. Both of those platforms were targeting sites not people. Sure, people are the ones who run the sites but the issue was the low quality business model, not the people being abusive.
Twitter needs to combat abuse. I can’t in my right conscience say that someone should be banned forever because they made one threatening tweet, as I do believe that people can change. I also believe that sometimes people need a bit of tough love.
How I’d Tackle It
So here’s how I’d tackle it for Twitter.
First offense where you are reported and Twitter’s systems (which probably need a human team to start with) verify that you were abusive (rape threats, calling someone a derogatory name, etc) gets you a 30 day ban. This is similar to Google’s 30-day manual penalty for link spam.
Second offense, you get a longer ban say 90 days. Google does this as well for more egregious types of spam.
Third offense, you’re gone. Three strikes and you’re out policy. The corollary to Google here is deindexation – you’re nowhere to be found. Though I think Twitter should go a step further and completely remove their account from the system. Google can’t do that because they don’t host sites, but Twitter does host accounts. They should be gone forever.
My bet is that implementing a system like this and being consistent in its application would go a long ways towards combatting spam and abuse on the site.
They could also take it a step further and use the inputs on muting/blocking to inform if a profile should be removed or its reach suppressed.
That could be another level – if you have a lot of people blocking you or muting you, have a system to figure out why this is happening. If you’re posting just links to articles with no engagement, lower the account’s reach to its followers. If you’re being abusive, go back to the three strikes policy.
Ultimately, Twitter isn’t going to be able to solve this issue until they get their public company messaging and metrics under control (there are people who are great at this). They have to move away from users being the driver of their stock price in order to really deal with this issue head on, but that does not mean that they shouldn’t start solving this problem right now.
To be clear, I love Twitter the platform. It’s given me a lot. And from time to time I’ll see something that gives me a glimmer of hope that they’re thinking about these things, but when I see Jack talking about edit buttons instead of abuse I tend to get sad.
I’m hoping they pull through.
Disclosure: my wife and I own some Facebook stock and I make a living off the back of Google’s platform.