An algorithmic Twitter is a very different tool

Twitter_512x512I think it’s hard to estimate just how greatly the Twitter experience would change if, as reported, the social network shifts to an algorithm-driven Timeline.

While everyone has reacted very strongly to this, particularly in the PR and media worlds, I’m trying to take a more cautious approach for the time being, at least until we know more. After all, this may be something we can turn off if we opt to, in which case it impacts my personal experience not a whit.

Despite the lack of insights as to how this might work (though we can take some guesses based on how Twitter has started showing random Favorited tweets in people’s feeds) there are some areas I’ve been thinking about and which I’ve seen others begin to speculate on:

For brand publishers the impact could be huge, either positively or negatively. On the upside, brand tweets usually see higher engagement levels than those from individuals. So this could be a good thing, surfacing those updates in more people’s feeds and increasing their exposure. But, as we’ve all seen from Facebook in the last year or so, algorithms can be manipulated by the networks that put them in place to their own ends and based on their priorities as a company, not based on the best interests of the audience.

Misinformation would take much longer to disapprove. Think about the last time there was a major news event and how things went down on Twitter. There was the initial blast of sketchy facts followed by a period where details became more and more clear until the real story was clear. But how much longer did those initial inaccurate tweets still appear in your timeline as people just catching up on things shared the news? It’s been my experience that the initial, inaccurate stuff sees much higher engagement than the later corrections. So if items are being ranked on engagement there’s the possibility the garbage will be given priority over the later updates. That’s a real problem that Facebook faces now and it would be a shame to see Twitter go down this same route.

Time-shifting would denigrate the value of the real-time feed. Again, think about what the current Facebook experience is: Your Newsfeed is probably a mix of posts ranging anywhere from the previous hour to five days ago. So instead of getting the real-time experience of what’s happening *now* Twitter would become another platform that’s a random mix of what *has* happened. And that degrades one of the core components of Twitter.

The begging for Favs or RTs would get out of hand quickly, likely leading to some sort of crackdown on the practice, which would mean the value of those points in the algorithm would have to be thrown into question, making said algorithm just that much more mysterious. And the use of media – photos and videos – that usually create higher levels of engagement might have to be curtailed by the algorithm since it could be seen as gaming the system by publishers.

The biggest part of Twitter that would benefit from this is the “lean back experience.” So the people who would get the most out of this are those who follow mostly celebrities and stars of some sort. Celebrity tweets are off-the-charts in terms of engagement and are usually not very timely, making them perfect for an algorithm-based format. So the extreme casual user is the biggest beneficiary of this, which is entirely the point.

In short this is a can of worms that I don’t really think Twitter wants to open. I understand it’s doing this out of a desire to make the experience more friendly for new or light users. Which is why this needs to be either an opt-in or opt-out experience, likely the latter since a new user isn’t going to know they need to opt-out of an impure feed.

Twitter is messy. While it’s a corporately-owned, centrally-managed tool the “how” of Twitter has almost always been in the hands of the users. I use Lists, other people don’t. Someone else is really into hashtag tracking, that’s not my thing. I use Tweetdeck, other people only use Twitter on their mobile devices. Innovations that are core to the experience like hashtags, @mentions and so on have all bubbled up from the user base, not from the company itself until they co-opted them and made them into feature sets. I get that it’s exactly that messiness that keeps some people at arms’ length, sticking with following Britney Spears and Zac Efron and that’s it. But it’s also exactly that messiness that makes it such a wonderful place. Yeah, you miss a lot if you’re not paying attention, but that has to be OK. Corporate decision-making cannot be held captive to some people’s fear of missing out.

As I said on Twitter (of course) the other day, it’s never been a place where I was concerned about finding news I *needed.* I have RSS for that and have actually just gone and added subscriptions to some sites I felt I was missing out on. It’s been a place where I found news that other people thought was interesting enough. If I missed something, well, them’s the breaks. But I like weird, messy Twitter a lot more than I like Facebook, where a group of engineers in California are making judgements about what should or shouldn’t be important to me without my input at all. That’s a level of control I’m not ready to give up, which is why I like RSS feeds so much. And for those who complain that they can’t manage everyone they’re following on Twitter, the solution isn’t this. It’s the Unfollow button, which is the best mechanism at hand right now to help you define your own signal-to-noise comfort level.

There’s no accountability in deciding decency on social networks

Here’s the key graf in this story about how much control companies like Twitter and Facebook have over, beyond things like a News Feed or other algorithm, what we see when we use their social networks:

So why am I so uncomfortable with this? Because it’s not clear what’s too vile to host. And, even more, because Twitter and YouTube are among a tiny group of giant companies with greater and greater power—and less and less accountability—over what we read, hear, and watch online.

That’s a role that used to be played by an editor within a news organization. And that editor was not only challenged by their staff but also held accountable by an ombudsman, publisher or other figure. And the public could register their outrage or support through various channels about the decisions being made.

Twitter and Facebook though have continued to straddle the line between being a media company, where roles like those would make sense, and dumb networks where anyone can do anything without interference. So we’re left with the question of who is making these decisions? The Twitter CEO? A Facebook publishing manager? Who? There’s little to no transparency in the process, leaving the public in the dark about what will or won’t cross some unknown line at any given time.

These and other social networks need something that goes beyond a standard Terms of Service that, realistically, seems to be selectively applied. There needs to be some sort of public editor in charge of decency who can address each individual instance where a story is too indecent to be hosted on its servers and speak to *why* it crosses that threshold. Someone needs to answer the questions being asked and be accountable not just to management but critics, the audience and others.

On a separate but related topic, this is the second best graf in the story:

The first way we users of Internet services can re-decentralize is to create—and make use of—our own home base online. In practical terms, this means getting your own domain name and creating, at a minimum, a blog where you establish your own identity. The page you think is yours at LinkedIn, Tumblr, Instagram (Facebook), or any of the other centralized services is emphatically not truly your own; it’s theirs.

In other words: Keep the web weird.

The same people saying click-bait is a problem are the ones clicking on that bait

facebook_logo.pngI was going to write something really profound about Facebook’s announcement that it was cracking down on “click bait” type stories, penalizing them the News Feed algorithm based on a number of factors, but then Dave Coustan wrote this and why bother. See his post for how this is likely to impact brand and other publishers.

The question I keep asking, though, is how did Facebook ultimately decide this as a problem it needed to address? It says these types of posts were resulting in a steady amount of negative feedback, both on pages themselves and in a survey, where people said these types of “…and you won’t believe what happened next” articles weren’t very good.

But then why do they so frequently appear in people’s News Feed? I can understand why people can feel like they’ve been the victim of a bait-and-switch, but then why take the steps of engaging with that post in a such a way that similar stories show up more regularly?

This is, to me, a bit like surveys where people tell a news organization they want more hard-hitting stories about international news but then tune in only when one celebrity gets in a traffic accident while driving naked. Surveys are notoriously bad ways of gauging actual behavior and the changes now being made based on those results are now going to impact a lot of publishing programs.

There’s beauty in Twitter’s unfiltered stream

Both Mathew Ingram at GigaOm and Charlie Warzel at Buzzfeed (there are others but those are the biggest ones I’ve seen) have pieces about the stark contrast between what people are seeing on Facebook and what people are seeing on Twitter over the last week or so.

The gist of this is that Facebook has been filled with videos of people taking the Ice Bucket Challenge, a stunt designed to raise awareness of and get people to donate to ALS charities while Twitter has been the place to turn to for updates about what’s happening in Ferguson, MO as well as more recent stories about the beheading of a journalist in Syria and more.

I won’t belabor the point but instead encourage everyone to read those two pieces linked above. But I will take a moment and reiterate how the difference is largely – almost solely – because one uses an algorithm to decide what to display (Facebook) while the other is the unfiltered stream (Twitter).

While Twitter’s lack of algorithm may not mean someone sees everything – a lot is missed in the time we’re looking away – it does mean they see a lot more. And it’s easier for people to catch up because there’s often someone who’s X amount of time behind the news and who’s sharing things now, a bit later than when it was breaking.

As Twitter makes more and more rumblings about introducing some sort of algorithm to the Timeline (something that isn’t necessarily bad…as long as they either make it an opt-in feature or something I can easily opt out of) I hope they’re looking at all the commentary about this issue and that it’s giving them pause. While they can muck around with officially making Favorites something that appears in your Timeline in addition to Retweets all they want, the beauty is in the stream. It’s messy, it’s imperfect and it’s often wrong. But it’s also a magnificent example of a wonderful, flowing public conversation. And it’s *much* more important to have this sort of free-flow of ideas and news than be subject to some system’s idea of what is or isn’t important.

Two Tragedies, Two Opportunities to Reconsider Social Media Publishing

People were stunned Tuesday afternoon when news broke that Robin Williams had, by his own hand, passed away. That includes myself. Williams was a huge talent and an incredibly loved and influential comedian, actor and person.

As the afternoon more and more people in the “Social Media Marketing” wing of Twitter particularly came out with their opinion that, in the wake of the tragedy, brands should pause their social publishing programs. While I certainly see their point, I disagreed. Williams’ death was certainly sudden and Twitter was filled with an outpouring of emotion over it, but in my opinion it didn’t rise to the level where brands appear incredibly insensitive and tone-deaf with their scheduled updates. There was nothing inappropriate about continuing to publish at that time like there is when there’s a shooting at a school, a bomb goes off somewhere in the U.S. or something similarly massive happens. It’s hard calculus, and I certainly respected differing opinions, but this didn’t meet the necessary criteria to make that recommendation.

(Note that if we paused every time a bomb goes off elsewhere in the world no brands would ever tweet again until the second coming of Christ himself. We also differentiate between shootings. 12 people could be killed in Chicago over this weekend and every brand in America would continue to publish without a second thought.)

After watching on Twitter as events unfolded in Ferguson, MO last night, though, I’m increasingly of the opinion that the situation *there* does cross that threshold.

Think about it: Right now in an American city – one just 275 miles from my front door – there are police confronting unarmed citizens with riot gear, including long-range rifles, tear gas and more. That’s really happening. Journalists are being taken to prison, peaceful protestors are being tear-gassed and more.



(image via @chicagotribune)

Imagine if brands took a stand and said, “In light of the actions happening in Ferguson, MO we are taking a break from our social media marketing. When we feel the situation has been resolved, or there’s a clear path that’s being taken to that resolution, we will resume regular posts. Until then, our thoughts are with the people of Ferguson.”

Now that might sound like crazy talk. Take a stand on a public issue that has little to nothing to do with a company’s business? That’s nuts and this shouldn’t impact marketing one way or the other, right? But that’s exactly what brands and companies are doing everyday with issues like birth control, same-sex marriage and a host of other social issues.

Just like taking a stand on same-sex marriage shows companies attempting to convey human emotions and take a stand against what is the status quo, doing so around the events of Ferguson would be a showing that they’re not blind to the events around them, events that are, on many levels, hard to believe.

I’ll admit I’m not ready to make this call myself just yet. It’s so outside the norm that I’m still hesitant to recommend what I’ve outlined above. But it’s something that’s growing and growing in the back of my mind and, if things aren’t resolved there soon, I may work up the nerve to at least try to lead by example.

Facebook Save and the needed next step in read-it-later apps

Facebook introduced Save last week, their own version of a “read it later” service that allows people to save interesting stories for reading at a later time. The idea being that while they’re scanning Facebook they may see something that looks interesting but they don’t have time right then to click through to the story or to fully digest it, so they need to save it for later, when they presumably have more time. This puts them in competition with existing services like Pocket, Instapaper and a handful of others.


When I first read about the service it struck me as a solution in need of a problem. That’s largely because of not only how I see people using Facebook, which is in large chunks of time so they can sufficiently scan their Newsfeed, but also in how Facebook has spent so many calories recently positioning itself as part of the ongoing, current conversation. They want to be the place where people are talking about what’s on TV, what events are happening and so on, so anything that is geared toward time-shifting seems inconsistent and a bit odd.

Going back to the first point, “save for later” functionality has in my experience always been something that’s been sought by the power-user set and not necessarily by the mass audience, who just want to go online to check email and check for updates from friends, family and the brands who they’re hoping share coupons with them. That’s part of the reason why RSS, which by its very nature was all about time-shifted viewing and saving deep reading for a later time, never caught on, because the desire to do so just wasn’t there in a mass audience. They didn’t need to do that because they had their bookmarked pages and were fine with that workflow. And it’s part of the reason why the average web user hasn’t even heard of Pocket et al, because that’s not how they’re consuming content online.

So it’s not clear who Facebook Save is really meant for. And the best guess I can make is that it’s intended to be something that brand publishers are supposed to applaud because it dangles the possibility that while someone may not click through to a story now, they may do so later.

But metrics for Facebook Save were not part of the announcement and it’s unclear as to whether this is something that will be introduced or added on down the road. Without it, though, it will be difficult for publishers to truly measure how and to what extent people are using this tool.

Facebook Save is, though, not unique in this manner. To the best of my knowledge none of the big save-for-later apps offer any sort of metrics for publishers. And those numbers are absolutely necessary for publishers in the same way that “day plus seven” ratings numbers have increasingly become necessary for TV networks. Let’s walk through why:

I, as a brand publisher, post X story on Facebook today. And I can see in my site analytics that views to that page/post spiked in the period immediately following that, knowing that links on Facebook (or Twitter or anywhere else) have a very short half-life. There may be a few moderate bumps down the road but generally after a certain period of time you’re going to see those numbers flat-line.

But now factor in save-for-later functionality, whether it’s within the Facebook framework or in something like Pocket, which integrates with Feedly, Digg Reader, Twitter’s mobile app and more. If I, as publisher, can start to attribute traffic to those apps then I have that much more complete a picture of my readers. And it’s not just that, since apps like Pocket allow for reading completely within the app, without sending any traffic to the source site. So if these apps were to work with publishers to show how many people were saving stories from their domains, how many articles they were saving, where they save the article from (Twitter, RSS, Facebook etc), how long they saved the story before they read it and so on, then all of a sudden publishers are swimming in additional reader data that can help drive strategy on a number of levels.

It also opens up a world of possibilities for the apps themselves, as well as their business models specifically. Imagine if apps were able to sell ads to publishers that looked like recommended reads, giving readers the option to either save it for later reading or to go ahead and visit the site directly. There’s value in both options.

There’s also this from Om Malik:

Those are great ideas that would, yes, require some development work. But honestly a network that’s based on similar reading habits is of much more use to me than some other social networks. It’s basically what Google Reader Shared Items was before it was so brutally ignored and killed. It’s easy to see scenarios where I say “Ooo…X Friend would like this story, I’ll message it to them” within the Pocket – or other app – framework.

On an even more basic level, these apps need to more fully embrace a variety of networks. For instance, it would be nice to be able to choose LinkedIn, Flipboard or other networks to share a story from Pocket instead of just Twitter or Facebook. (Note that I don’t use Buffer, which is an option within Pocket. Not because I don’t want to, I’ve just never taken the time to set something like this up and it’s for an even thinner slice of the audience than users of save-for-later apps.)

Better metrics and data would serve the save-for-later market well. Since Facebook is now treading into these waters in a serious (even if it might be a drastic misreading of their audience) manner it’s a good time for the existing players to start mining new areas to prove their value to readers and publishers, both important stakeholders in how the tools they produce get used.

The homepage for all things Thilk


Get every new post delivered to your Inbox.

Join 2,604 other followers

%d bloggers like this: