Does Sir Malcolm Rifkind actually understand the internet? As chair of the Intelligence and Security Committee he has pointed the finger at an un-named internet giant, later revealed to be Facebook, for failing to report a threat by Lee Rigby’s killers.
There seems to be an assumption, in the coverage of this, that Facebook actually knew about the threat.
Hello? There are around 125 million posts a day on Facebook. It is totally unrealistic to expect Facebook to monitor all those posts and report suspicious ones. It’s not a question of hunting for a needle in a haystack. It is like searching for a molecule in an ocean. They would need a massive department and mega-equipment to do such a thing.
On Today this morning, Richard Barrett, former Director of Global Counter Terrorism Operations for MI6, expresed a similar view. Simon Hughes confirmed that such firms as Facebook are under no current legal duty to report such things.
And by the way, the government has an entire sophisticated department monitoring internet traffic. They didn’t pick up on the Rigby killers, apparently. So why expect Facebook to do so?
Update 13:01: The Guardian have now posted the quotes from Richard Barrett, former MI6 Director of Global Terrorism Operations, from this morning’s Today:
Richard Barrett, a former counter-terrorism chief at MI5 and MI6, said that he doubted the capacity of Facebook to sift through the volume of content handled by the website each day, as well as that of the security services to deal with the amount of information that would be referred to them if an obligation was placed on internet companies.
“Facebook has about one and a third billion users and about five billion posts a day so clearly on a worldwide basis it would be almost impossible to deal with the amount of stuff that was referred,” he told BBC Radio 4’s Today programme.
“And even in the United Kingdom there are about 25 million users of Facebook and so let’s say possibly about 125m posts a day. And even if you take out all the pictures of kittens which were put up you’d still be left with an awful lot to go through and then quite a percentage of those perhaps would be passed on for the police or security services to look at. So it would be an enormous task, I think.”
* Paul Walter is a Liberal Democrat activist and member of the Liberal Democrat Voice team. He blogs at Liberal Burblings.
20 Comments
Facebook does though cut corners as it admitted to me – it has bare minimum staff to monitor any reports and I doubt if many of these have even basic education
It is a money making transphobic organisation (and I can back that up) that is careless about what is posted
Yes it is impossible to police every post but it makes little attempt to respond to concerns either
I am reminded that when from 1869 onwards postal services (beginning with Austria) began to introduce postcards (the novel idea of sending unsealed messages through the post) some postal administrations printed a disclaimer on the cards they sold stating that that the postal service had no legal responsibility for the messages written on the cards. It was never the post offices’ job to read the millions of cards sent and to check whether their messages were lawful. Similarly, when the telephone was invented, it was never the responsibility of telephone companies (and state telephone operators) to listen in to conversations to check whether anything unlawful was taking place. Facebook and the internet generally are no different. Providing a communication service does not imply having a law enforcement duty. It would be like selling a car and being held responsible for its driver’s (and drivers’!) driving offences.
As to Facebook in particular, there is also an issue of practicality when looking out for suspicious messages: it is used in more than 100 different languages with dozens of different alphabets and any language can turn up on anyone’s Facebook account in any country of the world. The implied notion that you would only need to monitor its communications in English would be parochial in the extreme.
so it’s just alchemy then that in amongst zillions of posts Facebook manage to put up all those adverts on my page aimed at women of 54 who have exactly the same online spending habits as me? Some stuff they can track, but some they can’t. Seems a tad inconsistent.
Does Paul Walter actually understand the internet?
>It’s not a question of hunting for a needle in a haystack.
>It is like searching for a molecule in an ocean.
No, it really isn’t; you’re comparing an ocean with a digital system designed for searching! You find me that molecule in the ocean, I’ll find you whatever language patterns you want in a database, and we’ll see who gets a result fastest! Obviously, searching databases is a process that’s happening all the time, everywhere, but I don’t believe you have a solid, repeatable process for finding any molecule in the ocean – that would be astonishing and not similar to a load of regex/sql. The needle/haystack convention is what’s normally taught regarding search algorithms, perhaps your molecule/ocean will catch on one day, but that seems unlikely.
>They would need a massive department and
>mega-equipment to do such a thing.
Top assessment – you should supply IT equipment! They have all of this mega-technology already, they currently use it to read your posts and target massive mega-advertising at you! If third party spammers can make a living scraping Facebook with libcurl, why is it so outlandish to expect the firm that runs the database to scrape it for terrorism or social threats?
Finally, I’d like to point out that “sophistication” is relative – the most sophisticated systems for these sorts of jobs are usually owned by the organisation that spent the most money on it. As such, neither the government nor Facebook are best placed to do this job, but Google has some truly amazing technology in the arena.
…does ChrisB actually understand the internet? etc, etc
ChrisB
“you should supply IT equipment! ”
I do
LOL! Is it mega-massive IT equipment for finding molecules in oceans?!! 🙂
Oh dear this is embarrassing. Yes. 🙂
It’s not like a molecule in an ocean, but it may be like a lost boat in an ocean. People are much more complicated than molecules. And it’s not equipment that’s needed, it’s software, because it’s perfectly possible nowadays fro software to detect patterns and meaning. And it’s not just the software, its the legal backing that’s needed too. Which is where many of the issues reside.
In the meantime, facebook are “withholding the contents of five accounts set up by Fusilier Lee Rigby’s murderer Michael Adebowale”.
http://www.telegraph.co.uk/news/uknews/terrorism-in-the-uk/11255039/Lee-Rigby-report-Facebook-still-withholding-details-of-five-accounts-held-by-Michael-Adebowale.html
If they do monitor I hope they are better at it than Google. My other half is interested in the history of art. Google keeps sending her adverts for painters and decorators!
Brilliant post ChrisB.
It’s worth noting this quote from the ISC report into the event leading up to Rigby’s murder:
“The Home Office explained the particular issue US CSPs have raised, that: complying with [the Regulation of Investigatory Powers Act] would leave US companies in breach of US legislation (including the Wiretap Act in relation to lawful interception)”
The Home Office have been maintaining for many years that RIPA and other UK legislation can be used to force foreign companies to hand over data. The reply to such requests is inevitably the same as the response I always give when foreign corporations and governments have requested information from me: A short but polite “No”.
The security services were already monitoring Lee Rigby’s killers and we are told almost daily about how sophisticated their internet tracking is, plus not so long ago a bill was rushed through parliament to retro-actively make legal activities they had carried out illegally for years. But some how they didn’t look at Facebook! And this all comes to light as the Home Secretary is preparing to introduce new monitoring measures. Really!
“They have all of this mega-technology already, they currently use it to read your posts and target massive mega-advertising at you!”
Yes, and that’s easy. Just find a few things in posts that indicate what someone might want to buy, target a few ads accordingly, and you’re done. You don’t have to find everything in the posts, you don’t have to get all your guesses right.
Now think about searching for words that prove someone is a jihadi. You’ll get a flood of “Murder on the Orient Express”, “I could die for a burger”, and “My feet are killing me, and I need to go to the mosque and pray”. Now pick out the genuine jihadi from this flood of near-hits, and don’t make any mistakes. Not so simple.
Gail Bones
“so it’s just alchemy then that in amongst zillions of posts Facebook manage to put up all those adverts on my page aimed at women of 54 who have exactly the same online spending habits as me?”
David Allan pretty much comers it but one additional point. If your adds are wrong what happens, you don’t buy the product aimed at a 20 year old man . If you flag an incentive person who posts “just walked back from the mosque, my fewer are killing me” then what happens?
You start trawling their private info, they may be added to a suspect list which could impact on their lives.
Technology does not have a cure for everything. Assuming it does poses a real risk to us.
This “shoot the messenger” stuff about Facebook is a shabby attempt to shift responsibility from the agencies appointed to keep us safe. Time after time in such cases it transpires that there was knowledge of the individuals involved but preventative action was not taken in time. In this particular case the report stresses that there are limited resources and if a specific case is judged to need more urgent attention that can only be addressed by shifting resource from other cases.
Does that not make the very point that GCHQ and other resources should be concentrated on cases of real potential. The constant pressure to (in effect) regard the entire population as suspects and then try to narrow it down from there must eat up resources. We are frequently hearing of people being shocked to find there are “files” on them – simply because they have put their heads above the parapet in some way – whistleblowers and the like. I have no specialist knowledge in this area but it seems clear to me that much better focus would yield results.
“….And even if you take out all the pictures of kittens which were put up you’d still be left with an awful lot to go through..”
This irresponsible complacency could endanger the people of this great land. Has the former head of MI5 and MI6 lost his senses? Does he not realise the threat from Jihadi Kittens returning from Syria?
@David Allen
>Not so simple.
I don’t think you’re aware how far Google have developed semantic search – your examples are trivial if you’ve trained a neural network to spot the differences in these usages. It’s deeper linguistic twists such as compound zeugmas that exist over long phrases rather than single words that take a lot of strategising to achieve successful interpretation. There are countless examples and papers from the past decade that deal explicitly with the sorts of issues you raise. So, not simple but well within current technological bounds – Google will identify the difference in meaning between your examples right now, live as you type it in via its convolutional neural net. They’re now refining the same technology for images; an order of magnitude more difficult, but the process is the same – feeding data in and confirming or denying the accuracy of the semantic interpretation by the neural network (feed-forward). The last time I read up on this many of these processes were at about 95% accuracy across all languages, I expect they’ve improved since then (they’re not short of input).
Soon there will be only one company that can keep us “safe” and at that point all of this talk of governments and politics will be redundant. If Google can find the terrorists and GCHQ can’t, why would we need GCHQ? I think algorithmic governance will become a major political factor in the next decade and politicians have been very naive in their approach to these matters.
ChrisB
“Soon there will be only one company that can keep us “safe” and at that point all of this talk of governments and politics will be redundant. If Google can find the terrorists and GCHQ can’t, why would we need GCHQ? I think algorithmic governance will become a major political factor in the next decade and politicians have been very naive in their approach to these matters.”
I think you are rather over confident in technology.
I think this whole debate of whether Facebook could do this is sort of missing the point. I think the key question is ‘should’ Facebook be acting as our new ‘thought’ police.
Now, before anyone says ‘they have a moral duty to stop people being killed if they know it is going to happen’; well, let us think of this. Even if we take the assumption that such a duty exists (morally – which I would be willing to probably in most cases concede it does), there is a problem that Facebook did not know this would happen, and in 99.99999999999999% of cases, they are unlikely to know because it would be PURE CHANCE for them to find such posts, unless they are actively looking for such posts.
Which brings us back to the original questions, ‘do we really want Facebook actively looking through our messages and acting a new thought police for the Government?’
>I think you are rather over confident in technology.
No, just under-confident in politicians ability to legislate sensibly on it. I don’t think these are good shifts, I’d rather we supported our national security services rather than try to persuade external corporations to police their own systems.