Stats-based response to UK Tories' call for social media terrorism policing


#1

[Permalink]


#2

So Cameron wants businesses to take over the burden of detection work, and snitch on their customers. Also he expects businesses to devise the methods for deciding who to snitch on. Because small government.


#3

Lions are Everywhere!!


#4

Well which is worse? That one bad person goes free? Or thousands and thousands are oppressed by their government? It’s important to stop evil, you know!


#5

Another bloody stupid Cameron crack-pot idea. When will that man learn how things actually work?


#6

Ach, yes, the classic age-old dilemma, you know: with one hundred suspects, is it better to execute 99 innocents than to let the one guilty go free?


#7

He’s a politician. This modus operandi is his job description. No chance of it getting any better. :frowning:


#8

He knows how things work: he gets born into money because he deserves it; he goes to Eton and Oxford because he deserves to; he marries into more money because he deserves to; he becomes Prime Minister through knowing the right people (who he met at Eton and Oxford and his wedding) because he’s the right person; and when he gets kicked out of that job, he’ll get some lucrative directorship sinecures and lectureship gigs because he clearly is a superior person.


#9

Like the Trolley Problem - but with riot gear!


#10

Pedantically, the example is actually misleading about the statistical model. There is no reason why an algorithm needs to have equal miss and false alarm rates (in this case, .01). They typically don’t, and this is also typically a criterion on evidence that can be set by the decision maker based on the relative costs. In a case like this, the optimal evidence criterion would be much different, depending on the assumed costs of false alarms versus correct identifications. For example, if we are willing to reduce target identifications from 99.9% to 50% or 10% of terrorists in any given year, we might be able to reduce the number of false alarms to an acceptable level, and still produce a small enough set of messages that can be combed through by a human. In cases like ‘trial by jury’, we in theory try to err on reducing correct convictions to avoid false convictions, although the way it is set up, it is probably pretty biased against the defendant. When detecting cancer, the test-makers are willing to flag more false cases, to be sure fewer cancerous cases are missed. This may actually be part of why screening for colon and other cancers is not effective.


#11

Hopefully, in May…


#12

Cameron will still be PM. Lab-Con grand coalition to keep out pesky SNP and UKIP MPs.


#13

You and I and the boingboing commenteriat know this because we know about evidence-based science. The political class do not know about science or math, and so do not support evidence-based law-making. Instead they respond to socio-political pressure.

When the next terrorist act occurs, the Daily Mail and their ilk will demand wider-embracing laws to catch more suspects before they commit terrorist acts, and the political class will pass them — as more badly thought-out emergency legislation because terrorism. And they will work, because they will catch more people breaking those laws, which will prove that they work.


#14

I’m hoping Labour-Green, if we’re lucky.


#15

I almost prefer an earlier generation, when the rich and powerful believed that being born rich and powerful was part of God’s plan.

At least there was a plan. And a semblance of a greater good.

Now they just make it up as they go along. Out of self-interest.


#16

I’d be hard pressed to say anything else remotely nice sounding about it; but this strategy is about the most coherent synthesis you are likely to see when a neoliberal aversion to taxes and confidence that the state can’t do anything right collide with a good, old fashioned, enthusiasm for reactionary authoritarianism.

When you want repression; but you don’t want to pay for it and it’s an article of faith that the state can’t handle it as well as the private sector(despite G4S’ repeated attempts to prove otherwise), this is pretty much your option.


#17

We’d be much better off with good political bullshit detector’ software. Now THAT would be really useful.


#18

That’s easy. Aim a camera at the politico. Detect face. Detect lips moving.


#19

This topic was automatically closed after 5 days. New replies are no longer allowed.