#RIPTwitter: What An Algorithmic Timeline Means For Brands

Friday afternoon, Buzzfeed published an article claiming that Twitter plans to introduce an algorithmic timeline as soon as this week, prompting such brouhaha that the hashtag #RIPTwitter trended throughout much of the weekend. Enraged users argued that it would bury small accounts and new users, favor brand over user content, be antithetical to the freedom and transparency of Twitter itself and, ultimately, make Twitter effectively indistinguishable from Facebook. Zeynep Tufecsi, Assistant Professor at UNC School of Information And Library Science, put it this way:

 

tweet1

While these reports have not been confirmed by Twitter, they haven’t exactly been denied either. A series of tweetsfrom CEO Jack Dorsey on Sunday afternoon began with the following:

Screen Shot 2016-02-08 at 11.31.35 AM

The addition of “next week”—those apparently superfluous 11 characters—keeps the door open, but it’s important to look at other signals as well. The introduction of the “While You Were Away” feature last year could be interpreted as a trial run for an algorithmic timeline, as it is, in Twitter’s own words, “determined by engagement and other factors.”

While Twitter hasn’t specified what those other factors are, contenders could be signals such as follower count, popularity of keywords or hashtags, whether you’ve retweeted or favorited that person’s content, your connections to that person, and other attributes. Numerous reports that an algorithmic timeline is currently being tested add credence to the idea that something is likely afoot. For the purposes of this post, I’m not going to speculate any further on the “will-they-or-won’t-they?”, but it is important to think about why Twitter would make such a change and what might happen if it does. Instead, we’ll look at the impact on three critical constituencies: 1) users, 2) brands and 3) researchers.

Why Would Twitter Do This?

The simple answer is user growth. Twitter’s vulnerability (especially in comparison to Facebook) has been amply documented, and the company needs to make itself more attractive to advertisers by increasing the size of its user base and its stickiness. Filtering content by high engagement “and other factors” would arguably increase the signal-to-noise ratio, thereby appealing more to new users and therefore advertisers. There is plenty to say about 1)how this might work and 2) whether it would work, but that’s for another day.

tweet2

  1. Impact to Users

User control is critical

The first question is whether an algorithmic timeline would be optional or not. If it’s optional, Twitter will likely design the user experience to encourage people to use it. Why? See above. It’s unlikely to be a complete switchover, although the hope would be that users prefer the algorithm over the real-time timeline and use that more frequently, which would theoretically teach the algorithm more about what they care about and surface ever more relevant content.

Organic reach and viewability are no longer a guarantee

As on Facebook, users who publish tweets would not automatically be assured that all their followers are able to see them. Of course, given the velocity of the timeline (especially for people who follow a lot of people), there is never an absolute guarantee, but the algorithm’s function is to favor some tweets over others, so organic reach would no longer be a reliable metric.

Potential to erode trust

Granted, the #RIPTwitter hashtag is by its nature an extremely biased sample, but one of the implicit themes in the tweets is an erosion of trust—to be an open exchange and a place where events of national and global significance (#ArabSpring, #BlackLivesMatter, #Ferguson) unfold organically in real time. While “trust” is difficult to measure, loyalty is not, and erosion of the user base, particularly the most influential users, would be dire.

Do users know what’s best for them?

Another of the themes in the #RIPTwitter backlash, on both sides, is the question as to whether users actually do know what is best for them. As this writer for BBC News accurately points out, “If [Jack Dorsey] looks at the history of Facebook, he will see that just about every innovation introduced by Mark Zuckerberg has met with dismay from existing users.”

A tweet from Bret Taylor, former Facebook CTO, echoes this thought from another angle: “Algorithmic feed was always the thing people said they didn’t want but demonstrated they did via every conceivable metric. It’s just better.” But not every user feels this way, or is likely to act on it. One user wrote, “I don’t want to see tweets YOU DECIDE I want to see. I’m a big girl and can navigate my feed all by myself.” Facebook has been able to navigate this balance, but it is unclear whether Twitter is in a position to do so.

  1. Impact to Brands

Uncertain insight metrics

One of the main reasons brands use Twitter is to discover insights from users. This is much more reliable (but never 100% certain) with a real-time feed. Some brands currently have access to the entire Twitter “firehose” via social analytics partners, and this would conceivably continue, but the sheer existence of an algorithmic timeline would change the nature of the feed itself. For example, some critics have expressed concern that some events that trend in a real-time timeline would no longer do so with an algorithmic one. If, for example, one of the ingredients in the algorithm is social influence, that possibility could be magnified, especially for less “influential” users.

Uncertain reach/viewability metrics

One of the challenges Facebook faces with advertisers relates to reach (or with video, “viewability”) metrics; that is, the ability to know how many people have theoretically seen a particular piece of content within a particular time frame. With a real-time feed based on follower counts, this is a relatively simple equation, even when adjusted for the fact that people don’t sit in front of their computers/mobile devices all day (yet!).

With an algorithm, however, there is no way to know exactly how many people saw a particular piece of content, which has made ad buying a bit of a delicate dance. This panel from the Clean Ads I/O Conference last year lays out these issues quite effectively in context of Facebook. Is Twitter ready to enter this fray with advertisers? I’m not sure they are; Facebook has the benefit of more than one billion daily active users to bolster its negotiating power.

Uncertain value proposition

While it seems logical on its face that an algorithmic feed would give Twitter some of the advantages of Facebook, the opposite may well be true. Twitter’s appeal is its real-time nature; replacing it with an algorithm undermines the platform’s main differentiator. At the same time, it would bring Twitter into more direct competition with Facebook, and with many fewer and less developed features to compete with. My take is that, as a brand manager, it still wouldn’t change the value proposition in Twitter’s favor. That requires a much higher level of innovation than Twitter has shown during the past several years.

  1. Impact to Researchers

My Big Boulder Initiative colleague Farida Vis (@flygirltwo) has written a must-read blog post for the London School of Economics blog, unpacking the potential impacts of feature changes on research and arguing for researchers to be “more data and platform resilient”. One brief example is as follows:

In a research context…feature changes can have all sorts of implications for data collection and analysis. Take for example a news context: would users ‘heart’ a tweet about a recent terrorist attack in the same way they might have ‘faved’ it for later reading or for use in a news story? The way in which users engage with these features should of course be taken on board in analysis, but is harder to do as changes are rolled out and happen during data collection and user communities don’t adopt new features as expected. Feature changes have long represented a significant methodological challenge to researchers.

This applies both to academic and business research, as feature changes, if not identified and adjusted for, can wreak havoc on any type of analysis.

All of this said, I’m personally ambivalent about the prospect of an algorithmic timeline on Twitter. I have used the service daily for nine years and consider myself a committed user. So many of the most powerful events and movements of the last decade have unfolded in 140-character bursts, educating me about issues and points of view I might never have been exposed to otherwise. That is irreplaceable, and I’m grateful for it.

As an analyst, however, I’m frustrated by the innovation slump that the company has been in. Twitter has assets unlike any other on the planet, and the company needs to be able to monetize that. Is an algorithmic feed the answer? I am skeptical, but acknowledge that all of this is theoretical until we see real news from the company. But one thing is certain: I really hope Twitter takes an unsparing look at the data from #RIPTwitter and uses it to make its next move one that brands, researchers and users alike will benefit from.