Data portability in online dating sooner than they think?

Dating Apps make money from attention & personal data

I have written a few times about disruption in online dating, heck its something which will be discussed at Mozilla Festival this year (tickets are available now).

But interestingly the EU’s General Data Protection Regulation may get in there ahead of any setup/network disruption. In the Guardian I saw a piece called Getting your data out of Tinder is really hard – but it shouldn’t be.

Its all about getting data back from Tinder (which remember is part of IAC/Match group)

…Duportail eventually got some of the rest of her data, but only on a voluntary basis, and only after she identified herself as a journalist. Her non-journalist friends who followed suit never got responses to similar requests.

Finally armed with the 800 pages she had clawed back from Tinder, Duportail wrote a story reflecting on her own relationship with her data, and the myopic view Tinder had of her love life. I feel her story helps bridge the chasm between those with information stored in the database and the architects behind it, providing much needed neutral common ground to democratically discuss power distributions in the digital economy.

Given the popularity of her story, and my overflowing inbox, I would say many agree. And indeed, you should expect more similar stories to be unearthed in the future because of the upcoming General Data Protection Regulation (GDPR). From May 2018, the new European-level regulation will come into force, claiming wider applicability – including on US-based companies, such as Tinder, processing the personal data of Europeans – and harmonising data protection and enforcement by “levelling up” protections for all European residents.

I know there is a lot of push back from the big American internet corps, but this is coming and the there is no way they can wriggle out of it?

…beyond the much older right of access, the true revolution of GDPR will come in the form of a new right for all European citizens: the right to portability.

It seems like such a small thing but actually it has the potential to be extremely disruptive. Heck its one of the things I wanted back in early 2011. Imagine all those new services which could act like brokers and enable choice! It could be standard to have the ability to export and import rich data sets like Attention profile markup language (APML).

I just wish we were staying in Europe, although the UK has agreed to take GDPR, thankfully! There was no way, if they were left on their own, this would ever come about; like it looks like it might.

Data portability in dating, we can only hope

From OnlineDatingPost

Speaking of sharing, Twitpic Blocks Posterous’ Import Tool; Out Come The Lawyers. Data portability isn’t something discussed in the dating industry very often in public. Dating sites sell profiles up and down the river every day, but that’s a big shhhh! topic. Everywhere else, data portability is the topic of the moment, and will be for many quarters to come.

While I’m glad the message is getting through, without some standards like APML your not going to see any portability in the dating field. Even who is one of the enlighten sites doesn’t have support for Openid, Oauth or anything like that. Supporting profile and data sharing or portability is going to be a major stumbling block for any dating site, simply because the revenue models rely upon data being held, locked down and processed by the company. This doesn’t fit with data portability right now.

Once they work out they can track people around the web using cookies and other techniques (which I assume will make there searches stronger and there for make for better matches) I’m sure you will start to see they at least providing openid if not oauth between partner sites.

Lightweight Attention Preference Markup

So this is the 2nd time I'm writing this because I forgot to save the entry when I upgraded the memory on my Dell. Yep 2gig of memory instead of 1gig now but still no decent Blogging tool for Linux. Wblogger and Ecto would have automaticlly saved the entry every few minutes or at least asked me what I should do with the unsaved entry before terminating and throwing my words to a black hole. Anyway enough moaning…

Previously I promised a couple of things in this entry

First up, I'm going to standardise some way of linking FOAF, OPML, OpenID and APML together. I expect I'll keep this very simple using the link element in (x)HTML or somehow combine this into a Hcard profile. Next up a APML microformat or APML lite for sure. I'll try it as I've been studying the others and the general methology of Microformats and I think it could be done. So I'll suggest it and draw up how it works and submit it for lots of review. I'm now exploring how to get APML out of Amarok and RSS Owl.

So how far have I got so far?

One : So I have linked all three (APML, FOAF and OpenID) together using links on my blog. So if you look at the source you will now see this. Which is cool but I think we can do better.

<link rel="openid.server" href=""/>
<link rel="openid.delegate" href=""/>
<link rel="meta" type="application/rdf+xml" title="FOAF" href=""/>
<link rel="meta" type="text+xml" title="APML"

When I say do better, I've been looking around a couple of things. First up is a better way to do the basic link element so it can be turned into a RDF triple later. It was found while looking at RDF/A examples which will be explained later.

When a meta or link is used within another meta or link, the internal triple has, as subject, the external triple. This is reification.

<link about="" rel="[cc:license]" href="">  <meta property="dc:date" content="2005-10-18" /> </link>

which yields:

[ rdf:subject <>; rdf:predicate cc:license ; rdf:object <> ] dc:date "2005-10-18".

Now I'm not that keen on the syntax, but its not over complex and I guess you could do something like this.

	<link about="." rel="[foaf/images/emoticons/silly.giferson]" href="">
	<meta property="apml:profile" content="" />
	<meta property="openid.server" content=""/>
	<meta property="openid.delegate" content=""/>

But I guess getting all those openID parsers to change now will be a nightmare, so to be honest I'm happy either way. But I think it does make sense to link everything in the HTML rather that rely on a OpenID parser to look at the HTML then find the URL for the FOAF file and then parse through that to find the Open ID url. Yes I already know you can put OpenID in FOAF thats why I'm saying its not a good idea, but there is no harm in having it in the FOAF optionally. Which is what I'm going to do, but I've recenly seen how out of date my FOAF file really is, so I'm going to try and update it soon. If anyone knows how to get FOAF out of Facebook, Flickr, Delicious, Linkedin, Dopplr, Upcoming, etc that would be useful. O'reilly's connections network use to allow for FOAF but somehwere along the line seems to have died or closed down, because I tried to find it and login, so I can at least start somewhere. So generally number one is done.

Two : So the huge challenge of building a Microformat for APML, so people can easily put in there preferences without building a very complex xml file. Because lets be honest, like RDF and other XML's this stuff was never meant to be built by humans. Also I like the idea of using standard HTML elements and attributes so people can instantly try this stuff out. I saw recently on the microformats blog that there is almost 450 million? examples of Microformats now and its growing everyday. Its not hard to see why when you consider how it is to try out some of them. For example adding a tag is as simple as adding another attribute to a link. Some of the other microformats are a little more tricky but generally with a example in front of most people they can work it out quickly. So whats the W3C's answer to Microformats? Well RDF/A which is a unified framework build around putting semantic meaning into HTML. A while ago it was meant to be for XHTML 2.0 but its been brough forward which is great news. Because the only other alternative seemed to be e-RDF which no one could work out if was royality free or not. Ok I have to admit I'm writing this entry over a couple of days. So I found my way on to the O'reilly connections network again. So you should be able to see my public view here. Anyway the point is that they already have FOAF, which makes my life slightly easier that starting from scratch again. Going back to APML, I'll try modeling it with RDF/A and see what happens. So far I think my plans is to keep the explict and implicit context and maybe attach it to a openID or unique ID. I'm not going to include stuff like the source because its too complex and not that relevent for a lightweight version of APML. I mean if you really want APML, just use APML. If you want something to indicate your preferences (< href="">beyond a link) in HTML, what I'm brewing up might just be right for you. I've also decided to call it LiteAPM, as in Lightweight Attention Preference Markup for now.

Three : Ok I'm not being funny but where the hell does Amarok store its configurations and database? I think I've found RSSOwl's basic configuration stuff but content i'm not so sure about yet. But then again I've not really tried really hard yet. I can't find a mention about Amarok anywhere. So I hit the web and found a way to pull almost anything I want out of Amarok via the command line. So honestly all I really need now is to learn how to program Perl or install something like XMLstarlet, and learn how to use stuff like the cron and unix pipes. Wow now I can do all that stuff I've been talking about for a long time. Stay tuned…

Comments [Comments]
Trackbacks [0]

APML as a lightweight Microformat?

My Ubuntu install is working again thanks to Jon Callas from PGP. So I can finally bring you this entry from the Airplane trip 2 days ago.

I've not blogged much of BarCampBrighton, but sitting on this flight to Boston with 2.5 hours of battery life left (54%) I'm now reflecting on a couple of my sessions

The first one was titled as Facebook is Dead, which I later changed to social network killer – reclaim your attention. The whole thing presentation went through reasons and technologies which mean the dead end, locked in social networks no needed to exist. However Jeremy Keith and Tantek had covered most of my points about Microformats and OpenID in a previous session so I skipped most of that stuff and got on to the juicy part about lifestreams and attention profiles. In a quick look, I showed off APML and suggested ways you could do FOAF+APML together to build up everything you needed for open social networking (as aposed to locked in like Linkedin and Facebook style). Oh also while on the differences I made it clear that social media sites like Flickr, Delicious, Bliptv, etc were not included in the locked in category and then started to rip the living piss out of (no link for this nasty social network). I think I put the slide up and asked How dumb do you have to be?. Anyway I'm going off topic here.

So many people are throwing APML around but as Jeremy Keith would say, it looks like a technology looking for a solution. So I came up with some perfect examples of APML including my killer which I thought would also attract the bigger audience who don't care about the technology of it.

First example : Tape it off the If you join tape it off the or tioti it will ask you again and again and again… what shows you like and do you also like these shows? Now this is fine but its tiresome although you can type in your shows and hope it finds them. But what you can't do very well is give shows a certain weighting. So for example I love lost and heroes but have also started watching the new prision break 3rd season although yes its jumped the shark years ago. Now I would like to indicate that although I watch it, I'm not really that much into it. Then I also want to be able to show my friends and maybe get less matches based on this preference. Whats evern better is if I can say I really dislike something. Right now you can give it one star or not add it but what incase I want to say, I watched that show and I bloody hate it? In APML you can set a negative value to things as you'll see later. In actual fact Tioti is going to support APML in the future so that battle is done already.

Second example : Dating sites. This I thought was my killer one but maybe not? So if you join a dating site you build this profile of you and sometimes that involves answering questions and filling in a profile of who yougr looking for as such. This is usually quite detailed stuff and time sucking stuff. So imagine what happens when your friend invites you to a new dating site where he or she have previously had a positive experience. Yep fill in the forms again and indicate your preferences again. Now with APML its robust enough to maybe describe a lot of what your saying in your description and profile. You should be able to log in with your Open ID which could link to your FOAF and APML too. And because you have the concept of personas in Open ID, you could select different FOAF and APML files for those who like to have different profile. Tantek who was in the room when this was announced pretty much slammed APML but liked my example and so suggested I check out some of things people have been doing with tagging their preferences. So I did and although quite robust I had that problem of setting amounts.

So for example if I tagged this quote below you will see I've choosen to tag certain words. But there's no indication of first, if there just tags (yes some class could be added to make it clear that these tags apply to me but in what way?) and secondly what in case I tag flying and writing? How would you know I much prefer the later?

Facebook hopes to expand on the service, one person says, using algorithms to learn how receptive a person might be to an ad based on readily available information about activities and interests of not just a user but also his friends.

In APML you can add values and that makes a huge difference. But maybe there is a way to have a lighter version of APML which is a microformat, so people can play with it right now.

third example : Targeted advertising So as you read before, Facebook are using your data to advertise at you. Here's the full thing.

Next year, Facebook hopes to expand on the service, one person says, using algorithms to learn how receptive a person might be to an ad based on readily available information about activities and interests of not just a user but also his friends — even if the user hasn't explicitly expressed interest in a given topic. Facebook could then target ads accordingly.

While Facebook plans to protect its users' privacy and possibly give them an option to keep certain information completely private, some Facebook users might rebel against the use of their personal information for the company's gain.

And the perceptions that targeted ads create can be as much of a problem as the reality. “Most people don't realize how targeting works; it becomes so good that even though it's anonymous, you feel like they know you,” says Rishad Tobaccowala, CEO of Publicis Groupe-owned consulting firm Denuo Group. However, he says Facebook needs to be careful in implementing any targeted ad system, lest loyal users “find it creepy.”

Chris Saad, is right to ask the question if users will put up with it. Also just making things private isn't the answer. What in case I had kids but didn't want to see adverts for nappies and babyware? I could hide that stuff but actually don't mind telling the world that I had kids, I just don't want the adverts. In APML i could specify my distaste and give it a nasty negative value which would indicate that I never want to see Nappies or Babyware ever. Of course Facebook or Google could over ride my APML but I wouldn't be happy, and they should be looking at my positive APML stuff anyway. I mean there's more chance I'll click a link saying latest Orange SPV phone here that latest Babyware or even in reality Latest iPhone offer for O2. And this is the killer thing, anyone who knows me will know that, even a browse of my blog will tell you this but to have it in machine readable form in APML has got be final straw. On my blog my APML is linked and its public. I'm saying advertise to me if your high on my interest list. Orange should be banging my door down every time I say I'm looking for a new phone but there not wise to all this yet. Google and Facebook are.

Forth and Fifth examples : LastFM and Preferences. By now you should have got the main points and can see how having a APML output of your favorate tunes in Lastfm could be useful. One example I didn't think about at the time was moving between media players. I moved from iTunes to Winamp to Madman to Amarok. There was no way to take the ratings I had build up for my music collection but now I should be able to look at lastfm and pull out decent enough scores. On the preference front, APML could be used for more that just internet applications. So yes if Amarok supported APML that would be amazing. I'm actually looking into this because Amarok keeps a rating of all music based on how many times you heard it and how much you scored it. It also keeps this information is a SQLlite DB and the rating range is between 0 and 100 which makes it easy to put into APML. Yep I might need some help from some Perl or Python people soon. Although I've learned Gnome has its own scripting language like how Applescript works for OSX. So I might be able to pull something off that way.

Right so what happens next? First up, I'm going to standardise some way of linking FOAF, OPML, OpenID and APML together. I expect I'll keep this very simple using the link element in (x)HTML or somehow combine this into a Hcard profile. Next up a APML microformat or APML lite for sure. I'll try it as I've been studying the others and the general methology of Microformats and I think it could be done. So I'll suggest it and draw up how it works and submit it for lots of review. I'm now exploring how to get APML out of Amarok and RSS Owl. If anyone wants to help with these, give me a shout. APML is awesome and I know everyones waiting for a 1.0 release (including Tioti, Google, Bloglines) but honestly give it a try now you won't regret it.

Comments [Comments]
Trackbacks [0]

Why I love the idea of APML

APML support so far

I decided to split up my posts about the girl geekdinner because something happened later when we got to the pub in Victoria afterwards.

Walid from was showing me some of the new features there planning. Obviously these are not to be repeated so I won't. But we got talking about the Trusted places taste tester and Walid pointed out a site I've never seen before called Imagini. Now how we got on to that subject is about profiling. I was suggesting to Walid it would be great if you make the profiling data available to the user so they could tweak it or share it. Glyn asked about the business motivations for doing so. I didn't really have a answer except it would be very cool.

So why?

Well imagini tries to map out who you are by asking you about 13 questions. Its results are poor and very general. But worst still is once you've done all that work, you get rewarded with a widget, some facts about yourself according to them, some travel sites you might like and being added to their facewall. The author calls it VisualDNA, theres lots more about VisualDNA including this part which talks about the reasoning behind it.

Did you know that businesses around the globe spend a staggering $18 billion per year on market research, trying to work out better ways of understanding what we all want? On top of this, about another $350 billion is spent every year advertising to persuade us to buy what’s been produced and available…

We think that this is totally outdated and simply not a sustainable way to carry on. It just makes sense that the future must be about producing less whilst meeting peoples needs more. We believe that the changing way in which we are all using the internet will make this possible by enabling people to get together and share information about what they like, want and need.

Our view is that the way to start assisting this process is to open up a completely new method of communication – a language that everyone who can see can interact with and understand – a language of images that enables people to understand each other in a different way.

The reason we have chosen images as a way of doing this is because about 90% of the way we all communicate is non-verbal. This 90% is made up of all sorts of different components that include many visual aspects such as who we look, act and behave.


This may sound cool but I'm left thinking, what else is it for me?. Now imagine it created a APML (Attention Profile Markup Language) file along with everything else. Then that would be something special.

This got me thinking too, what if other more established places like Trustedplaces, Last.FM, etc also gave away a APML file as part of the profile of each user?

One of the things I loved about APML is the Implicit Data (U-AR) and Explicit Data (I-AM) elements. You can just imagine how simple it would be to output APML from something Last.FM. (whats below isn't true APML markup, just my lazy json like writing)

Implicit (U-AR) {
concept{ Ferry Corsten = 0.87 }
concept{ Armin Van Buuren = 0.90 }
concept{ Sugar Babes = 0.1 }
concept{ Lemonhead = 0.00001 }

Anyway thinking about Glyns question about the business angle, I still don't quite have an answer except to say I've been following Steve Gibsons Security Now which recently has been talking about multifactor authentication.

  1. Something you know
  2. Something you have
  3. Something you are
  4. Someone you know

Well I was thinking APML could be useful for 1 and 3 but started thinking about a 5th factor. Something you know about someone. So a question could be does friend1 prefer ferry corsten, Armin, sugar babes or lemonhead? Maybe? or Maybe not?

Anyway I look forward to seeing more applications and services using APML or something like it. I think there's business reason behind APML but I can't put my finger on it right now. Hopefully someone like Trusted places gets it before Digg who just annouced something similar to trustedplaces.

Comments [Comments]
Trackbacks [0]

Particls – now you can all pay attention…

Particls on my desktop

The private alpha ends today. Yep the guys from Faraday Media have made Particls available to anyone who wants it. Go get it now.

For those who don't know Particls is an extensible attention platform. It learns what you like to consume and gives you more of that. I have been using it for quite some time now and have found it very useful.

  • For users: Particls is a filtered news reader or widget that learns what you care about and alerts you to important news and information while you work. More at
  • For bloggers and site owners: Particls allows bloggers and site owners to create a custom version of the application. Particls will share revenue with partners. More at
  • For developers: Particls is freely extensible by developers. Reach into corporate databases and web APIs to grab and display data in new and interesting ways. More at
  • How much is it: Particls is a free download with some ads. Later, an ad-free Pro version will be available for a small subscription fee. It is free for Partners to create custom versions.

So Particls is the biggest step forward in the debate over attention. Some of the scenarios people have talked about can be played out in Particls. For example if Particls knows what your browsing about, it can throw up an alert from a site owner suggesting a 20% discount if the person buys that item they were searching for on right now. And thats just the start of things.

I once outlined a scenario where Particls is looking at your Microsoft Money account and whats in your Amazon wishlist. It notices you always get paid on the 28th of the month. So through clever logic pops up alerts with discounts for some of your items on your wishlist when you have enough money to pay for it.

This is quite scary but possible. And raises the issue of people taking control of their attention data. Which is where APML fits in perfectly. One of the things which always impressed me with Particls was the ability to look at the result of their I-AM/U-AR engine in XML and adjust it accordingly. This means you can just erase a large section of your personal attention data without too much hassle. It also means you can import from something else like another attention engine or your keywords from your lifestream for example.

So enough chatter, you can download it for the PC here or check out options for the Mac while they develop their native mac version.

Comments [Comments]
Trackbacks [0]

Plumbing for the next web at Xtech 2007

I have uploaded my presentation, pipelines: plumbing for the next web fresh from the first day of Xtech 2007 today to Slideshare.

The general view is that the presentation went down well and made sense. However I think people really wanted to see something which worked instead of slideware.

Comments [Comments]
Trackbacks [0]