Flattr algorithm

With the public release of Flattr 2.0 we also published new versions of our extensions. There are quite a few updates but the biggest one is the second iteration of our automatic Flattr algorithm. It has greatly improved from the first version so here is everything you need to know about the new one.

From the user testing we did (and our own experience) with the first version of the algorithm there were a lot of things we did right. But as always there were also some use-cases that didn’t feel good enough and others that were totally missed. So we went back to the drawing board to see how we could improve it, and we did. The algorithm now has two main ways of flattring content, here is how each works…

 

Attention-based flattrs

These flattrs are made towards specific pieces of content, in other words a specific page. It’s measured on a daily basis and flattrs are rewarded for different thresholds of attention to the content.

Different types of content are consumed in different ways and should therefore be rewarded in different ways. We are planning to implement support for many types of content but to start we have support for two – “Video” content and “Text” (the rest of the web). The key difference is that video needs much more attention to get flattrs and they are also spread over a longer time.

Textual content gets it’s first flattr after about 45 seconds (as long as you are active in your browser) and then receives up to 5 flattrs for the next 3 minutes. The return on flattrs is increased (the later flattrs are easier to receive) because we want to encourage good, long pieces of content compared to splitting articles into multiple pages (a normal method to generate more ad revenue).

Video on the other hand has significantly higher thresholds, first flattr happens after a couple of minutes and then it tops-out on 7 flattrs an hour later. Flattrs are also rewarded with diminishing return for videos. For the opposite reason of textual content, longer videos mostly have “lower value per minute” and we want that to be reflected.

The flattrs are made towards the specific URL and you will find the full URL in your Flattr dashboard.

 

Visit based flattrs

This type of flattr did not exist in the previous version of the algorithm and it has been introduced for two main reasons.

Firstly, we wanted to cover websites, content types and consumption patterns that were not covered previously. To give an example, previously a webcomic you returned to when there is new comic every other day or once a week would not receive any attention-based flattrs because you wouldn’t stay long enough, but returning to the site shows that it’s a site you care about. For this we now have a mechanic that flattrs the site itself if you visit it at least 3 times during one month, with at least one day between the visits.

Then we have web pages that have shorter pieces of content, but still give you value. A good example here would be your local newspaper. It might not have long articles, but you stay on their site and read multiple stories in a shorter timespan. In this case, you stay long enough for us to see that it’s not a site you were tricked into visiting or one where you try to find something, but you didn’t. For these pages you need to visit at least 4 pages (that you stay on for more than a few seconds) over a total duration of a minimum of a few minutes, to generate one flattr towards the site itself.

The Second reason is because the biggest mobile platform is Android, most android users use Chrome. Most people also use Chrome on their computer. If you are signed in to your Google account on both devices (something that most people are) the browser history is synced between the devices. That means we automatically will generate flattrs from visits done on your mobile phone. Pretty smart, right?

This is the second version of the algorithm and we hope it shows our dedication to making it as good and as accurate as we possibly can.

Got questions or feedback about the algorithm? Then let us know!

 

Find out about joining Flattr as a creator or a contributor.

6 thoughts on “The New Flattr Algorithm: How does it work?

  1. This sounds like (and has been) a great improvement to the previous algorithm. Next for some tough questions:

    “[…] a mechanic that flattrs the site itself if you visit it at least 3 times during one month, with at least one day between the visits.”

    So daily comics are excluded? Poor Dilbert and friends. Most webcomis I follow publish one–three strips per week, so I guess it should work for most cases.

    I’m curious about synced history. I’m logged in with Firefox Sync across multiple devices where I’ve installed the Flattr extension on each one and all sync their history to each other (four PCs and one Android device). As far as I’m aware, each Flattr client works entirely independent of each other. How is this handled? (I’m presuming badly.)

    Have you considered sourcing a user’s YouTube history as well? I use the YouTube app on Android to play videos through a Chromecast. device These views/attention aren’t recorded in any browser history, just my YouTube history. Their API includes how long the video is and how far into a user watched a video so it should be usable for Flattr.

    Also, why I’ve noticed that you consistently refers to Flattr tokens as “a flattr” or “flattrs” here on the blog (but less consistently Flattr/flattr on the main website). Why dilute and generalize your fairly new trademark? They’re tokens that give recipients a claim to a share of a future payment. I’m not sure what to call them exactly. I’ve followed your lead and called it “flattr” with a lowercase-f in my Flattr fee calculator. “A flattr” isn’t really a self-explanatory term. Should I call them “Flattr tokens” instead?

  2. What is going on with Flattrs on reddit sites? Does it just flattr Reddit the company itself? What about Tumblr, same story?

    Also, for all the sites in the Dashboard that have been listed as flattr’d, does that mean that they have opted in? That is a requirement on the server’s side, right?

  3. Matt Chellew, there is a switch on the Flattr.com dashboard page that says “Only show flattrs with receiver” (they mean to say “recipient”, but that is another matter.) Toggle that and it will filter out any URLs that you’ve flattr’ed but have not yet opted in to receiving funding through Flattr.

    You can Flattr any thing (meaning URL) on any website. Flattr integrates with some websites like Flickr, Twitch, Vimeo, GitHub, Tubmlr, YouTube, etc. where individual creators can link their accounts on these services to their Flattr account and receive Flattrs to content they post on these sites. (Open the extension popout while you’re on a content platform website like YouTube and it will identify recognized websites with a “Content go to creators” label underneath the domain name.) Reddit and Imgur doesn’t have this tight integration, so any funds would go to Reddit Inc. (if they ever opted-in to Flattr). I’d recommend you Flattr all the things all the time, and then go through your Dashboard once a month to weed out anything you don’t want to support. Flattr’ing reddit right now just signals to the Flattr team that they should work on integrating Reddit accounts with Flattr accounts.

  4. Hey Daniel, you’re right – each Flattr client does work independently for now. This is something we hope to address and we’ll be sure to keep you posted when there’s a solution here. In terms of the idea about sourcing YouTube history – we have thought about this and have looked into if it can be used, however it contains items just after a few seconds of watching, so it’s problematic, also to get access to that would require the contributor to connect their youtube account too, so it’s a bit more of a hurdle.

    We generally always use flattr in lowercase apart from when we talk about our company name, so the way you’ve referred to on your fee calculator is correct. Really like your fee calculator, by the way :)

  5. Hey Misses, if you accessed YouTube via Chrome (while logged in) on your mobile device then this would work, however not through the YouTube app. We are looking into solutions here but there are a few obstacles associated to this that we need to find solutions to. Let us know if you have any other questions!

Leave a Reply

Your email address will not be published. Required fields are marked *