This afternoon we have a widely reported government petition that has reached 3,250,000 (when I last looked) it's a petition that's probably clutching at straws when stacked against an official referendum, However it gives politicians an excuse if they want to take a different route and for that reason it's worth continuing with.
This petition was initially started by a member of the English Democrats at the point when the leave campaign appeared likely to lose to provide cover for a second referendum down the line. That it has come back to bite them in the arse must have gone down like a bucket of cold sick
However the "bright" lights of the leave campaign have all come out do denounce the poll with claims that it was Fraudulently created. The Sherlock Holmes and Watson of this investigation appear to be Louise Mensch and Andre Walker Theres only one problem, neither of them appear to know what they are talking about
Firstly let me say there is some evidence of some fraud in the non UK figures, the Vatican City total for one being many times the population of the city. Until the overall count hit 2 million there was very littls sign of bot activity, apart from the aforementioned Vatican City result. in fact the counts of the various nations seemed to closely mirror the numbers of the UK expat population in those countries, however since then there has been a rise in external votes to the UK and several small territories have drifted out of line, which is probably the result of 4chan script and bot attacks, but it's obvious and something the database people will sort out tomorrow morning when they get back from the weekend. Chopping fake results from a database is a relatively trivial exercise, but usually needs some hands on effort.
The UK Internal results however are a different matter entirely from all signs visible from the publicly available data there is no evidence that there is fraud. If we look at the figures on a constituency by constituency basis, the numbers of signatories in each constituency appears to be in proportion to the number of remain votes this makes a scripted attack very difficult.
If you're going to write an attack script and for it to work effectively you need to have it run invisibly, otherwise the database moderators will pick this up Firstly, those fakes you are making signing the petition need to blend in with the normal population. If you produce a purely random postcode, then you will end up with the Bot produced signatures spread evenly across the constituencies, this will be instantly obvious, however this hasnt happened, the distribution appears the same as the referendum remain results
to get the data to appear as it has Those who signed up have to be from constituencies where people mostly voted against, now this is more difficult than it might first appear. The constituency record appears to be decided by the postcode that is entered on a UK entry, so a random bot would need to pick random UK postcodes distributed across constituencies in such a way that it produces an undetectable distribution the same as the remain vote discovered on Friday morning.
so this is quite a major piece of data analysis, then followed by some fairly complex coding fitting in with a bit of complex data translation between postcodes and constituencies. This isn't a small job to be done by a handful of bot writers in their bedrooms, and it's not just a case of plugging a minor detail into a per-prepared hacking script
This is not the way that the 4Chan bots mentioned by Louise and Andre work Their process is to produce an obviously flawed result so that it is instantly obvious that they have done it and so the Vatican City result might be the result of their fuckery, but the vast majority not. In fact if you look at the data, less than 4% of the signatures received are international so seem possible to be part of the Fraudulent group. Of those susceptible international locations, only about 12% of the 4% appear to possibly be fraudulent,and those seem so obvious as to be easily excluded on Monday morning when people get back into work after the weekend.
A second point that it is claimed led to fraud is that the
internal counter for UK signatures appeared to freeze at a number just
over 300,000 the only problem with this is that individual constituency
counters then didn't stop. anyone with a shred of knowledge doesn't look
at that and say Conspiracy!, they instead think broken code, or someone
has forgotten to run the correct script to update that counter. This
fault has since been noticed and updated, but it doesn't stop this
spurious figure being run about as "Evidence"
Too many journalists and politicians seem to think their "I can write a webpage" translates into "I understand data handling, data security, database error checking and network security" when really they seem to lack basic understanding of the technicalities and are more Clouseau and Dreyfus rather than Holmes and Watson.
Update: the Faulty Vatican City data has been removed, the count for there is back down to 70. Well done parliaments data wranglers
No comments:
Post a Comment