Graph Search, Facebook’s new frontier

Photos I like

When Facebook first started hitting the mass public conscious, maybe about 2008/2009, I read a blog post that was making the argument that the CIA (or was it the FBI?), was behind some of the funding. Not all of the funding, as far as the CIA/FBI was concerned it really didn’t want to be funding all of it. The argument was that they would want just enough of a foot in the door to be able to see the data, all those self-declared social networks. I read through the claims of money men and dodgy connections, people sitting on this board, and that board, boards of companies apparently known to have connections with other companies that were apparently known to be CIA/FBI fronts. It wasn’t massively convincing to be honest. The reason I kept reading to the end was that, it seemed to me, that if the FBI or the CIA didn’t have a piece of this new wonderful soon to be all conquering data thing, then they certainly should have, considering their raison d’etre.

Today we are all much more aware of what the social revolution has done in terms of generating personal data on private servers, and seemingly very comfortable about that too. We are also much more aware that the governments of the world simply aren’t shy of asking for whatever data they want, Google publishes a report every year and it’s not the tinpot republics, it’s us, it’s the UK and the US and France et al. In short these are paradigmatic shifts that have effected over a very short time span. So, even if the Feds had been buying influence all that time ago, it appears today, that they have decided to just ask, with some success, for access instead.

The point of all this being that the social layer contains very interesting, some might say, valuable information. Indeed this is the reason why Facebook was valued so strongly at IPO. There was a belief that as the owners of the biggest most specific and unique personal data set in the world they would surely find a way to leverage that data to sell effective ads to advertisers.

So, far that simply hasn’t happened. The utility of the data, as made available to advertisers, for targeting their ads simply hasn’t translated into enough sales, and the various innovations that Facebook have experimented with have also struggled to deliver performances that make them a mainstay of the modern advertising plan.

At the heart of this challenge is a truth which I believe is bad news indeed for display advertising on Facebook’s platform. Facebook is an essentially personal platform. Indeed it is the most personal mass adopted platform we know of. Because small ads (web display ads are essentially small in comparison to say TV, let alone big outdoor posters) generate smaller relative rates, and because Facebook usage is heading to mobile (smaller again), we would need to dominate the Facebook experience with advertising to get anywhere near to the revenue requirements of Facebook’s IPO.

Furthermore the more we work out how to get advertising into the Facebook feed (and I’m not even going to account for whether the ads will work or not yet), then by definition, the less personal the Facebook feed becomes. The less personal the Facebook feed is, the less time people will spend interacting with it.

Damned if you do, and damned if you don’t.

Faced with such a challenge what would you do? The answer it seems is to build a search engine.

Broadly speaking there are 3 business models for the web that are relevant to social properties.

  • The classic fee model, you pay for a service, a simple quid pro quo.
  • The freemium model, where scale is built with a functional free entry level product, which is monetised against a small sub section of users that need greater functionality and are willing to pay
  • Search advertising, the search model delivers a highly actionable form of consumer intent

It’s very difficult to retrospectively enforce a fee on a previously free service, such a move would destroy Facebook’s biggest asset, the size of its user base. Similarly the freemium model could also have a negative effect on the inherent user base, although exactly how they rolled out such a plan would increase or decrease such a risk. With such a large user base small but high frequency payments can be substantial when scaled up. Indeed Facebook is in the middle of trialling such a product with the pay for contact concept where you can pay to be able to access people that are not actually in your network.

The 3rd option, which of course has Google as a proud exemplar, would seem to be the longer term goal of the recently launched graph search. If there was a business model to try to emulate then surely it would be Google’s. To do that, you first have to build a search engine.

Facebook have been fairly clear about what they are up to actually, even if the ‘evidence’ we have are conversations from promotional interviews.

Steven Levy’s piece in Wired is revealing, in several ways.

““Right now our user experience on Facebook is a little passive,” he says. “Graph Search is a way to ask a specific question, to express an intent in some way. And of course an advertiser would want to target that intent. That’s what search ads are for.””

“Graph Search will be improved based on how people actually use it. So Facebook plans a slow introduction, limiting the initial rollout to a small number of users. Zuckerberg’s expectation is that by the time it becomes available to millions it will be considerably improved.”

So we know at least 2 things.

The delivery of the full vision of graph search will take some time. Amongst other things it doesn’t for example currently cover status updates or posts and it will only deliver real time location once it is ported to mobile (it is desktop only at the moment).

Secondly we know that an ad search product will arrive at some point. This alone is reason enough to take a long hard look at graph search.

It’s on an invite only beta roll out so the chances are that most people, today, will only have second hand access to what graph search is and what it looks like. I do not have access, I don’t even have a live Facebook account, I ditched the whole thing some time ago convinced that their lackadaisical attitude to privacy was something I didn’t want to get into bed with.

Fortunately, as it’s a search product it’s quite easy to conceptualise just from looking at a few screen grabs. The open web has been very kind and provided us with just that, courtesy of Tom Scott at his site

Tom’s site is somewhat comedic and dramatic, although he is not actively seeking to make the arguments for those concerned with privacy. Nonetheless, almost as a result of its flippant nature, we can get a really great flavour of exactly what graph search might be able to deliver. Some are mildly amusing.

EDL and Curry

Others really are anything but amusing

Islam men and women

The results returned by graph search are people, or more accurately people’s facebook profiles, pictures or places, and where the search query can not be satisfied with these data sets (that exist on Facebook’s servers courtesy of your Facebook sharing), Bing has been included. And indeed the inclusion of Bing is somewhat revealing. It suggests that the goal is to become more than just a tool for finding your old photos on Facebook. It clearly wants to become a serious destination for search functionality, the kind of functionality that transcends Facebook’s borders.

This, by the way, reveals the biggest challenge on graph search’s horizon, being useful outside of the confines of what Facebook already is (a warm pigpen where ‘friends’ share things). Much of what is being demonstrated, the bits that actually seem like aha moments are inherently personal, such as ‘photos of my friends before 1990’ which would return a host of delightfully nostalgic pictures of you and your friends from 20+ years ago. I think that’s pretty cool actually, even if it doesn’t stretch the tool outside of Facebook’s borders.

If you read a number of the reviews, promo pieces and reports there are a smallish number of classes of search that get mentioned time and again.

  • Photos of you and your friends
  • Finding a romantic interest / stalking (depending on whether or not the piece is positive or negative)
  • Organising a party of some kind
  • Recruitment
  • Where your friends like to eat
  • Where your friends have been
  • What music are my friends listening to
  • What books are my friends reading

Out of this entire list I only see one thing that is actually stretching Facebook into new directions and that is recruitment. Indeed it appears as though it might be quite a powerful tool for employers seeking to poach talent. In Zuckerberg’s words.

“One of my favorite queries is recruiting,” Zuckerberg said. “Let’s say we’re trying to find engineers at Google who are friends of engineers at Facebook.” He typed in the query and found, not surprisingly, that there were lots of people who met those criteria. Each one was represented by a little rectangle of information — their profile photo, along with snippets of key information like where they went to school, where they live, and the names of the mutual friends……..

…….“The good thing is that there’s people at the end of these connections,” Zuckerberg said. “You can find the right people or content page and then send a message.”

Everything else on that list just seems to make the existing Facebook experience a little more precise. I can’t tell you how many of my friends claim that they only use Facebook, ‘these days’ , to organise parties (or more likely small drinks at the pub, trips to the cinema – we aren’t getting any younger. I also think some of them are fibbing, by the way. I’m pretty sure, some of them at least, are still addicted to their feeds – who is doing what and where). Graph search certainly promises to add greater precision to that process although I remain unconvinced why one needs that precision.

Think of inviting friends out for a meal. Using graph search for such a task seems somewhat  dissonant to me, it forces me to ask myself all kinds of questions about the nature of friendship. From my perspective I already know, without a search tool, all the people who I would want to spend a friendly evening with in a restaurant and which cuisines they have problems with.

But, my problems here are irrelevant, I am an avowed non user anyway, and I’m getting on a bit, I’m not the target audience. That said there is still a concern here, regardless of my surly grump. How does this extend Facebook into new territory?

Let’s move back a step. Why does that even matter? Why is new territory important?

It matters because we are interested in whether or not Facebook can make a profit from graph search. Can it live up to its IPO valuation? And eventually, although the timescale will be long (myspace after all is still with us) will Facebook survive?

I have 2 questions that are focussing my thinking here.

1. Can graph search deliver actionable intent

2. Can graph search survive privacy requirements

The answer to number 1 is yes, but in a somewhat limited fashion in comparison to something like Google. This is why the issue of extending outside of Facebook’s current borders is so important. Google is outside of Facebook’s borders.

One commenter observed that they could now only invite friends that enjoy Thai food, to a prospective meal at the local Thai restaurant. There is clearly an expressed intention here. He is clearly intending to take a group of people to a Thai restaurant. For that reason it’s not beyond the bounds of feasability to suggest that a well placed ad might move him from restaurant A to restaurant B.

But here’s the thing, he’s probably already got a restaurant in mind. This whole idea of a social night out starts somewhere right?

  • He knows a great restaurant he wants to share with his friends
  • He has a desire to spend some time with his friends, but he’s not sure where to go

It’s rare to contact some people for a night out, and then ask them where to go. And whereas, clearly graph search means you wouldn’t have to ask them, the search tool would answer the query for you, I remain unconvinced that people think in these patterns.

I don’t really buy that he neither knows who he wants to dine with, or where he wants to dine with those people that he hasn’t yet identified. By the time he gets to Facebook he surely has one of those ideas, if not both, pretty much understood? Isn’t that just the way people mostly work? (Mostly is an important quantification here, there are always exceptions, the search ad business however is one of volume, not exceptions).

Maybe he has a desire to experiment with Thai food? But I can’t really imagine him doing that with a large party of people.

The one category where I can see graph search being useful to him is if he has his eye on someone ‘special’ and needs the cover of a big night out. And that takes us back to romance /stalking. I will expand on this area when I look at privacy in a moment. But for now it is a great example of how graph search extends precision into the existing facebook experience, rather than outside of it.

I think many people will still use Google to look at reviews of the various Thai options available. There is another paradox at the heart of graph search. Zuckerberg wants this tool to aid in the deepening of Facebook’s original mission, to be a tool of discovery. And why not, it’s a rather nice mission after all.

Again using the Thai restaurant as an example, if you know your friends well (the people you see most days, your family, people you have real relationships with) then you already know where their favourite Thai restaurant is, you’ve probably already been there with them. But then there’s the other kind of friend on Facebook, the old school mate, the colleague you met twice, the stranger you agreed to friend but who you are too embarrassed to unfriend and the person whose name you recognise but you’re not sure where from.

Ok, those people might well be able to guide you towards the discovery of a wonderful new Thai restaurant, but if you’re asking strangers anyway (I don’t care for the nomenclature, but these people are strangers as far as restaurant recommendations go) then why wouldn’t you go to a tool that allows you to query the opinions of all the strangers on the internet? Well, you would, quite frankly. It’s called Google and it would guide you to websites that have been built specifically to aid you in the discovery of new Thai restaurants.

And take a breath.

Will there be actionable intent? There might be some, but enough to be a business to rival Google (or to even get close)? No I don’t see it.

The second question, can graph search survive polite notions of privacy, is also a bit of a busted flush in my opinion. Let me explain why.

It would help to redefine the terms a little. Privacy, as strictly defined by the social paradigm, isn’t really the concern. Facebook has been clear that graph search will not surface data that could not already be surfaced via existing interfaces within the tool as it stands today.

Even putting aside many people’s concerns with the complexities of the privacy controls (even Zuckerberg’s sister can’t get those right!), and even putting aside the fact that Facebook has changed the TOS so that you can no longer opt out of search en masse (which by the way tells you all you need to know about their intentions here. If they really wanted to reassure the sceptics that they respected your wishes this would be the easiest thing in the world to offer them, but they won’t), the problems you need to consider are obscurity and aggregation.

Obscurity, or rather the protection of obscurity is what is really at stake. I read about this intriguing perspective in a piece in the Atlantic by Woodrow Hartzog and Evan Selinger, from, respectively, Samford University and the Rochester Institute of Technology.

The principle can be stated simply. It is the recognition that although data is public, the fact of its obscurity (partially defined as being hard to get hold of) means that people are still protected from its being used nefariously. A lot of data is public but very rarely if ever accessed, largely because it is difficult to do so.

We need to add in 2 additional concepts to discuss the possible harms here. Firstly the idea of edge cases being the focus of harms. I’m all right Jack is not the hallmark of a decent society, there are already examples of young gay people being outed against their wishes by this destruction of obscurity.

Most people aren’t gay. Does this mean you should not be concerned? Of course it doesn’t.

The second principle, and the one that crystallises this issue for me is the power of data aggregation. The potential to release previously unavailable insight and actionable data from happily shared information by combining it with other happily shared information. Some aggregation will not cause any problems, but again the edge cases, where the insight generated was most certainly something that the original sharer did not want revealed could be deeply unsettling.

Furthermore data aggregation, via graph search, makes the whole concept more actionable to considerably larger swathes of the population. Take for example this from graph search.


Single women


I’m going to pull this out a little.

Single Women – Ok, sure why not it’s no secret hey, and you’re looking for that great partner to share your life with, why would you hide such a thing.

Who live nearby – Again, on its own who would worry about that. It’s a big city right?

Who are interested in men – well, it makes sense to be straight up about that, it’s certainly not a secret is it?

And like Getting drunk – mmmmh. When did I tell Facebook that? Maybe when I liked my friends party pictures… mmm I’m not sure. Anyway I guess it is true, so what’s the harm?

Ok, so I don’t need to truly spell out what is wrong with this situation. We can just acknowledge that 4 pieces of information, shared separately, for innocuous and non-threatening reasons when combined together paint a very different scenario.

This is where graph search takes things to another level, see the tools on the side of the page, filter and expand (refine and extend)? So, it also becomes very easy to look at pictures of these people (don’t want to waste your time right?) but also if you need some more information to find them you can filter the detail to close in more accurately. You might only know that they live near you, which might put them in a group with 250,000 other people. But once you know that they work at widget company incorporated, well that’s a different matter, now they are in a group of only 50 (a lot easier than a ¼ million to sort through) and you know where the office is.

Right, time to nail my colours to the mast, although if you’ve read this far I’m guessing you know which side I’m on.

I think graph search is a truly dreadful extension of the Facebook empire, that will not only strip away what small protection remains (data obscurity) but will also hasten the day when a truly disgraceful and unforgivable crime is committed, one that removes the technology world’s ability to operate within safe and sensible boundaries with regard to data collection and data usage.

I’ve long said the biggest threat to the data status quo is the potential of aggregation. I had images of unpleasant men with access to industrial data, running queries in secret moments snatched in downtime at their places of work. From the relatively innocuous farming of data about people on holiday, to target empty houses for burglars, to considerably more nefarious acts of blackmail and violence I have long considered there will be an occurrence, an edge case no doubt, but an occurrence that causes enough outrage to get effective laws put in place.

Laws put in place because of outrage rarely get put in place and get it right. This is not an outcome I want to see. I want us to stop foolishly pushing the boundaries.

I never once imagined that the tools to hurry this potential would be delivered to every facebook user on the planet. Never once.

And, not that this is the crucial vector here, but I can’t see this damn thing even making them any decent sort of money either. Regardless of reach, from an advertiser’s perspective we haven’t, here, even considered the issue of data accuracy.

I am not a data luddite. My career is all about using data for marketing, I am on G+, Quora, Twitter, Linkedin and this blog. I am not an advocate of pure anonymity as a default setting online (although it should be an option in many situations), nor do I decry the clear need to monetise the current crop of free internet services.

But, is a little sense too much to ask for? Of Mark Zuckerberg? Who once said…

“The days of you having a different image for your work friends or co-workers and for the other people you know are probably coming to an end pretty quickly,” he said in an interview for David Kirkpatrick’s 2010 book, “The Facebook Effect.” Facebook users have “one identity,” he said.

One man, who can’t now even claim that he didn’t know what might happen, has decided that that long standing and deeply necessary reality of humanity, the ability to wear 2 or more hats (to borrow from Sir Humphrey) should be abolished. It is beyond my comprehension that he has taken it upon himself to make this real. Others have written many words in order to insult Mark Zuckerberg, so I don’t need to, I shall just leave us with the rather sad thought that it clearly is too much to ask for, a little sense that is.

Clearly far too much to ask for.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s