Here is some useful life advise. If someone is trying to sell you something and their patter involves counting something you should probably walk away if they want to switch methodologies. Wanting to count two similar things in different ways when you intend to then make a comparison is pretty much the warning that a con is coming.
After the anti-choice Vigil For Life last week received press coverage that 'several thousand' had attended I wrote up a piece pointing out that counting the crowd from their own video gave a figure of a little over 1000. The methodology I based this on was that used by a pro-life blogger who goes under the interesting name of Thirsty Gargoyle. After the March for Choice in September he had made a big deal out of doing a count of 'every person' in a video and on that basis proclaiming that only 850 were present.
Now actually this is a poor methodology for counting crowds unless your intention is to get the smallest possible number. For lots of reasons a video taken at a particular moment in time may miss a lot of people. They may not have arrived, they may have left, they may be behind a lampost. This Gargoyle seems to have an interest in crowd counting methods so he would know this, or if not has certainly discovered since as the blog I'm going to deal with here links to crowd counting resources.
Pretty much the best way to count a crowd is to station counting teams at two points on the route and do a complete count at both points plus at the second question a sample as to whether they passed the other counting point. This not only gives you two accurate counts but a reasonable approximation of how many additional people should be included who only marched part of the route. Counts are seldom done this way because its a bit of effort and march organisers seldom have the spare capacity. I actually do this on some demonstrations if I'm not busy which has the advantage of meaning I have a reasonable idea of what a crowd of X hundred actually looks like.
But Thirsty Gargoyle wasn't at the March for Choice so instead he picked the method that would give the lowest result and claimed that 850 people were there.
All well and good but the weird thing was that when he wanted to argue that my count, using the same methodology, of the Rally for Life was too low he wanted to use another method entirely. By this alternative one you calculate the size of the area occupied, assert how densely people are packed in and from this calculate how many people were there. It's a method that tends to give larger numbers than video counts so perhaps you understand why he wanted to change methods.
It can give startlingly different results. If we counted the March for Choice using that method from the area taken up at the end point and his claimed comfortable density we'd have got an attendance estimate of 5,150, This was actually twice our approximate count of 2500 but no less than 6 times his 'count' of 850.
The video count method returns a good minimum number. A 'we can be sure that at least this many people were there' number. The area count method produces a good maximum number. A good 'we can be sure no more than this many were there' count. Trying to insist on using one method to count one protest and the second to count another when you intend to then compare the two numbers is the sort of trick you pull on fools.
This is probably a good moment to point out that while my video count of the rally for life had returned 1029 his area count returned an estimate of 7560. Those of you quick at maths will realise that the second number is 7 times the 1st. Close enough to 6:1 to suggest that if the same method had been followed for both events there wouldn't be enough disagreement about numbers to make this blog worthwhile - indeed within a reasonable error margin they would both have been of the save size and tiny in comparison with the huge angry mobilisations that followed news of the death of Savita.
But here we are. And in fact after every pro-life tweeter had failed to notice my count for most of the week suddenly 3 or 5 of them got very exited indeed about TG's 'refutation'. No less than 25 excited tweets were directed at the WSM twitter account in the space of an hour this evening after the blog appeared. It all felt a little pent up.
This in itself was a little curious as TG had decided to settle on claiming there were 7,000 there but all week various pro-life sites & tweeters have been proclaiming that there were 10,000 or indeed more than 10,000 there. Why would they be delighted with TG for telling them that 30% less people were there in a reply to something they were pretending not to notice is a bit of a puzzle. In most people's book that would be a loss rather than a gain. Worse still the PLC tweeters seemed eager to blame the other pro-ligfe groups for what TJ had referred to as an "extravagant claim" but then these organisations have a rather large pot of US funding they are competing for and it probably seemed like a good moment to quietly put the knife in.
But lets hold off a moment as I need to make something clear. The different methodology TG choose to use may have given him a figure of 7,000 people for Rally for Life but even so the implementation of it was badly flawed. In fact when you starts to examine it you are forced to make some more thousands vanish to wherever the previous 3000 he has already done in have got to.
Life is short and the hour is late so I don't want to spend a lot of time on this but quickly here is the problem.
His estimate of 7560 assumes a tightly packed crowd spread out at a constant density over a clear space of 20m by 130m. The first problem is that 20m, the supposed width of Molesworth street. A quick check suggests that it is two wide by a meter - that makes sense as the outline he provides appears to include some roof space. But that is actually a minor problem. If you have been on Molesworth street or indeed if you just want to check on google satellite view you'll spot a bigger problem. Almost all the houses have deep basements in front of them, the sort with spiked fences to stop people climbing in. They are the width of a average car which means thats about 2m off that width on both sides of the road.
Ouch, suddenly TG has lost 25% of the space where he wanted to have the missing thousands standing. That's almost 2000 gone right there. I don't know how he could have missed such an obvious problem with his calculations.
The next problem is that not all of that remaining space was actually available. As I point out in my original post the organisers had left three rather large items on the street taking up a good chunk of space directly and indirectly - through creating poor visibility zones where no one would want to stand or more amusing 'stand here and get x-rayed' zones by the giant plasma cube of screens. I still want to know how much that thing cost and who paid for it.
TG wanted to poo poo any idea that these took up a significant amount of space but, in particular after you realise the 20m width is really 15m they do. The van is about 2*8m = 16m. The speakers platform about 10 * 6m = 60m. And that hell mouth of a video cube about another 6*8m - its truck mounted. Actually its a good deal bigger, Paul Reynolds picture (right) of it in operation for the night shows people standing 4m back from the screen and off the road entirely meaning that the screen and its X-ray exclusion zone gives us a giant 11m by 14m hole of 154 sqm.
So the actual space we have is about 130 *15 - (145+60+16). TGs 2600m shrinks down to 1729m. Now we could argue about that exact figure and really we'd probably need to go down with tape measures to resolve it precisely but give or take a bit there is a whole lot less standing room there than TG claims even presuming people are packed in all the way to the gates of the Dail (and the video shows they clearly were not).
So at the crowd density TG claims his new maximum attendance figure plummets to 1729 * 3 or 5187 people. Ouch. Youth Defence gave him a figure of 10,000 to open with and a consistent application of his own methodology has managed to half that number to 5000. With friends like those ..
It does I'm afraid get worse. Although TG has chosen to use a BBC article that suggests 3 people per square meter is 'comfortable' that is actually a pretty tightly packed crowd. Not as tight as a concert going 4 per m but Molesworth st wasn't exactly playing host to a Stone Roses reunion tour.
TG's been careful to mostly select crowd pictures from near the screen or stage where you'd expect people to be packed most densely but the one he provides of the Dawson st end of Molesworth st (at the top of this article) and indeed Youth Defence's own video of the Kildare street end show people that are widely spaced. You are looking at densities that are more like one or even two people per square meter and in the first case in the very area where TG needs there to be hidden 1000's. As I said in my original post it was easy enough to do a count for most of the scene but distance and angle of camera made the crowd half way down Molesworths st towards Dawson st tricky.
I've included a composite image to the right of photos photographer Paul Reynolds took at various points in the crowd that certainly demonstrate people are not packed in consistently. The messiness is all mine I hacked as many as I could fit into as small a space as possible without regards for his composition to illustrate this one point about density.
If in fact the densities we see in that top picture and in the video at Kildare street are typical we'd be giving a maximum crowd size guess at the 1760 mark, about 1 per meter. That's about 50% denser than a city centre pavement so I'm actually being a little generous here.
But lets not overstate things. I don't think any of the pictures show a significant section of crowd at a tight 3 per meter density. The US parks department would consider that a very packed in crowd. The photos I've seen suggest a wide variation from 1 person per 2 meters in the video at the Dail end to 2 people per meter in significant sections of the crowd near the screen & stage. Lacking an arial photo I'm going to suggest we won't ever know for sure but we can say that in terms of an area count that figure runs from a silly 880 to 3500.
I'm going to throw out that 880 because I managed to count more than that at 1029 and as I explained a video count sets your minimum number present just as your area count sets your maximum. So we can now say with some reasonableness that there were no less that 1029 people there and no more than 3500. Actually to be reasonable and take the low density sections of crowd into account that should really be more like no more than 2500 but its late and I'm not inclined to argue about that 1000. As I said in the original article certainly not 10k, 8k, several thousand or even with any sort of probability 4k. If TG can argue it down to 3k or 3.5k in a really convincing way (some pics of him out there with a tape measure would help) I might settle that high but my gut puts us around 2k - the first estimate I provided on the night.
I'll end as I ended the first article becuse the level of geekyness this blog contains threatens to distract us from the horrific reality women in Ireland continue to face.
I actually don't think the numbers mobilised have any importance as to whether women should have access to abortion in this country. My position is simple, if one women wants an abortion she should be able to access it regardless of how large or small a majority of people agree with her choice. That is what the pro-choice position means - the choice is for each women and not anyone else to make. However the political reality is that the arguments about numbers are going to be used to scare cowardly politicians so I'm not inclined to allow the crazy no abortion whatsoever forces that represent no more than 15% of the population these days away with anything.