Deified dicing of vCJD deaths unfairly fuels family fears
Scientists fear hidden epidemic of vCJD … after tests … extrapolated for the entire population... up to 3,800 positive. However... wrong to suggest because.. of the lack of information about the reliability... interesting observation... but we cannot work out what this means nor can we draw any conclusions… The most recent estimates based on computer predictions suggest the final size of the epidemic may not exceed 600 cases, but the analysis involves huge levels of uncertainty. To help resolve the issue, the Government ordered the testing of stored tissue samples collected anonymously as well as a much bigger survey of up to 100,000 tonsils and appendices removed over the next three years… investigated 14,964 tonsils and 1,739 appendices … reduced to 12,674 samples… findings of three positives … two of the three … looked quite different … can't make the assumption that those two positives represent positives ... There is still much to learn … he added.
More HIV-positive gay men are having unsafe sex … A third unaware that they are infected yet … engaging in unsafe sex … warned that risky behaviour … is a serious public health concern. … re-ignite the debate over whether the safe-sex message has lost its impact … The survey of 8,000 men … proportion engaging in risky behaviour … 30 per cent in 1996 to 42 per cent in 2000 … 1,206 men tested for HIV and about one in nine … antibody-positive … About a third - 43 … out of … 132 - … undiagnosed, … unprotected anal intercourse. HIV prevalence is high … many … undiagnosed ... public health concern … prevalence, anti-retroviral resistance … sexually transmitted infections … increasing … 10.9 per cent … in the survey were HIV-positive … Latest figures … 20 per cent increase in HIV during 2003…. 5,047 new cases … 4,204 in the previous year.
SADLY SCARILY SIMPLY SILLY
The top one has typical traits of another bureaucratic bandwagon rolling out another Aunt Sally of spiked statistics mirroring manipulations in the gemganic gangbang and climate conspiracy. In contrast to the apocalyptic alarm engendered by “epidemic” as headlined, the indicated incidence can’t be convincingly claimed to have shifted significantly from that estimated earlier especially with two of the three positives being merely maybe and even more with one in four samples rejected for degradation, puzzlingly so when prions are purportedly persistent. Simple sums (below) suffice to show its insignificance by converting the incidence of exceedingly sparse events like these pinhead fairies into everyday expectations from thrown dice, deified here as title allusion to Einstein’s reported remark because sadly whilst the statistics may be blind over the total population they certainly aren’t for individual people, not if eating habits and susceptibility play the part portrayed in conjectured causative connections.
The saddest significance of this study was cavalier contentment of its publicity promoting proponents to convey (more so on BBC R4’s Today) a cartoon of catastrophic consequence solely to amplify aspirations for an enormously extended evaluation at enormously extended expense and sod sympathy for poor souls fearful of the future having seen their siblings succumb to a dreadful demise having shared whatever genetic and dietary susceptibilities there may be for this nasty thing. Chasing fairies on pinheads would be fine if resource was infinite but it isn’t so much better to spend the money on functionally useful / helpful endeavours for the victims, not manipulating minuscule numbers in a nightmare lottery!
And as perspective on this navel-glazed number nibbling note that even on worst-case outcome this “epidemic” is forecast to “wipe out” in total about the same number of people as die on the roads every year, these deaths deemed endurable as inevitable consequence of living life and not impacting on everyday activities in any way, not even as second thoughts never mind doom-laden despair. Adopting an actuarial attitude and averaging the total over the 20-year indicative interval adopted in this arena starkly shows how this sad silliness really is no more than yet another Emperorial illusion!
By contrast the bottom one should be seen as significantly scary for its evidence of exponential expansion even two decades after that infamously haunting telly-spectre of graveyard gloom, still vivid for many with pubescent kids at that time. Extraordinary that its success (at least that’s the way its impact was played then and since but believable despite its Thatcherite origin) was never consolidated, especially with if being both corrective and cheaper than any amount of the symptom-suppressant palliatives that created complacency since cascaded into chaos and expert eyes still shut tight praying for inspiration on effective Aids-antagonists. The vista of a homosexual HIV holocaust cohabiting with an explosion in barely pubescent pregnancy certainly should be scary even for stoic fatalists never mind compensation cuckoos but there’s seldom a sign of even discreet bleating about these socio-sexual anomalies ~ notable exceptions being occasional outbursts by Tebbit reliving if not reviving his Thatcherite tenacity.
Me? ~ well I’m a meldrew so I blame it on correctness claptrap which has increasingly spawned seemingly ever stickier soaps that sadly serve as sole role models for far too many telly-traumatised tots to teens lacking any of the many reference rigidities enjoyed by earlier generations. I’m no Mary Whitehouse but when there’s little to nothing else as behavioural benchmarks it shouldn’t be surprising that morons mimic what they see most and the viewing figures show that most kids must see most of the pulped pap produced by populists seeking market share. Proclaiming innocence, asserting influence only as societal reflectors is a crap copout ~ everyone with an iota (and that’s a fearfully faded faculty) knows these perverted quislings really are the behavioural beamers defining declining standards, not simulating their devolution.
Anyway it’s manifest I don’t have a constructive commentary nor even a deconstructive demonstration to offer on this one beyond pleading for policies that heed the dangers declared in these data which have been trending one way dangerously for two decades without any sign of any active attention never mind intervention ~ extraordinarily in contrast to the cash colossus that climate cowboys created over the same interval for a conjecture house almost entirely constructed from computerised cards in contrast to decisive disease data in this despairing domain.
Now some simple sums showing how these vCJD stats don’t stack because their promotion was spiked to inflate fears so as to further the case for (much) more finance. I’ll keep this close to everyday experience so as to carry commonsense conviction in the conclusions. Consider coin-tossing for which there is chance ½ of getting either only heads or tails in two throws whereas the chance of getting only either heads or tails in three is ¼ so the chance of mixed outcomes (two of one, one of the other) is ¾. This illustration suffices to suggest that the threshold of surprise (or suspicion about the coin) lies somewhere between ½ and ¼, maybe 1/3 or so corresponding to chance of getting one out of two nominated numbers in a single dice throw. Nominating one and getting it (chance 1/6 of course) is surprising but nothing seems strange about nominating three numbers and getting one of them (chance ½ of course)
Adopting 1/3 as a ring of truth threshold for surprise / suspicion translates directly to significance in sampling for vCJD ~ getting a positive on 1/3 chance (ie against 2:1 odds) might well be seen as flagging significance for higher than expected incidence of disease. It gets a tad tougher connecting coins / dice precisely with prion positives but what’s needed is really no more than was done decades ago in A-level maths, although sadly since lapsed from firsthand familiarity judging from my own extensive experience of teaching university classes for a decade to the mid-90s. That’s sad because I taught the toughest topics to the brightest kids including goodly complements of triple A-graders (before they got degraded!). So with this experience of attenuated attention spans I’ll keep it quick but that means resorting to standard shorthand notation although nothing beyond potted primers like the old “Teach Yourself” things.
Taking a tail to be a “null” N (nothing) and a head to be a “point” P (positive) it is easy to check that coin-tossing outcomes over throws T are described by the binomial distribution (N+P)T with expansion
< NTP0T0/0!+Nt-1P1T1/1!+NT-2P2(1-T-1)T2/2!+NT-3P3(1-T-1)(1-2T-1)T3/3!+ etc >
where factorial (!) means 1!=1, 2!=2*1, 3!=3*2*1 etc and 0!=1. The coefficients of NT-tPt with t=0, 1, 2, 3, etc reproduce the chances of getting T-t Ns and t Ps. Confirm for yourself that T=2 returns one case with no Ps, two cases with an N and a P and one case with no Ns exactly as for coin-tossing outcomes from pairs of throws. Likewise T=3 returns one with no Ps, three with one P, three with one N, one with no Ns ~ eight cases in all and so ¼ chance of getting either all heads or all tails as noted earlier in making the point about commonsense conviction on significance.
To carry this description over to dice throwing for repetition chances of single pre-nominated numbers we need only to note there are five nulls for each point, so replacing N by 5N reflects this weighting and the coefficients above in turn must now be multiplied by 5T, 5T-1, 5T-2, 5T-3, etc, For T=1 there are 51=5 cases with no P for each case with 50=1 with no N, six cases in all and so a 1/6 chance of P. Similarly T=2 returns 52=25 cases with no P, 2*51=10 cases with one P and 50=1 case with two Ps, 36=62 cases in all and so a 1/36 chance of two Ps ~ again familiar from everyday experience.
The final step from coins through dice to prions is facilitated by setting P=RN with ratio as R=P/N and extracting the common factor N (no effect on the “distribution” over possible returns) before expanding the binomial in powers of P and exploiting the fact that R is very small so as to drop higher terms, so small indeed for prions that only a few terms suffice as we shall see. For coins then R=1 (no difference to numbers first thought of) whilst for dice R=1/5 so replacing P by RP means the terms in the expansion now have extra factors 1/50, 1/51, 1/52, 1/53, etc, As checksum, the pair-throw T=2 returns a distribution 1:2/5:1/25 for 2:1:0 Ns or 0:1:2 Ps, same as before. In fact we can already see how this thing is going for small R because even with R=1/5 commonsense rejects 2Ps as so unlikely as to be discounted for practical purposes ~ indeed reject it not just compared with 2Ns (1/25:1) but even against NP outcomes (1/25: 2/5) as a 1/10th rank outsider by our commonsense cut at 1/3rd. So for all practical purposes it makes sense to discriminate only the chance of “one or more” Ps as being effectively 2/5:1 meaning out of every 7 shots 5 return no Ps and 2 return one or more Ps, Indeed, for most practical purposes we may as well round this further and say that of every 6 shots 4 return no Ps and 2 return Ps and doing this means in dice-throwing terms here is about a 1/3rd chance of getting a P (any pre-nominated particular number) from two throws ~ true enough, in accordance with everyday experience and simply meaning that each shot is effectively independent so two shots each with a 1/6 chance so as to return a 1/3 composite chance.
In moving onto prions, there is one further trick that simplifies inspection of the odds ~ namely to note that the Ts as shown expanded <> above arise with same exponents as the Ps (now Rs) and can thus be bundled into a single quantity S=TP/N=TR, S standing for Signature. For all sensible scenarios with small S, T has to be big to recover anything at all and this means we can neglect all inverse powers in the boxed expansion above in which limit then the expansion is simply exponential in S or S0/0!+S1/1!+S2/2!+S3/3!+etc. Moreover small R means N can be adequately approximated by the population rather than population less positives, adequate even for practical purposes with dice as we saw above but immensely so here when S isn’t much bigger than unity so P/N isn’t much bigger than T-1 and we have already neglected everything in inverse powers of T.
Okay so that’s it for the gentle background and we’re now ready to try out some numbers. We may as well begin with what the pinhead fairies had in their survey, T=12K or so (upwards 16K with rejects) and accept for now their risk group is 10-30 year olds, say 18M or about 30% of 54M ~ 18M providing a close convenient basis for simple sums. The independent reference study by another group (probably also into apocalyptic amplification so we should be cautious on that score as well) projected 500 fatalities from “computer modelling” whatever that means in the absence of ground truth constraints ~ indeed, even with reality anchors, what’s the point in “computer modelling” when significance thresholds don’t warrant more than diced guidelines, never mind uncontrolled uncertainties than probably exceed significances anyway! Computer modelling is significant only as an attention attractor for plodding politicians, posturing politicos and their media mates, also bean baggers for future fortune from inflated funding.
Anyway, diatribes done, 500 in 18M would mean R-1=36K which with T=12K would mean S=1/3. Taking this as unit here is convenient because within all the overwhelming uncertainties there is no meaningful difference between exponential E1=2.7 and 23/2=2.8 so ES=23S/2 for practical purposes here. I know it seems arcane but this really is a neat trick in handling the numbers generated from integer multiples of S=1/3, doing all mentally without resorting even to a calculator! I’m old-fashioned ~ I’m old enough to recall slide-rules ~ I’m meldrewismic after all! As a hint on where the dancing with fairies people reside their “epidemic” number was approaching 4k fatalities, 8 times higher meaning an 8-fold higher R or an R-1 of 4-5K so we expect relevant S<6/3, not higher ~ all about unity or less so the exponential approximation stands.
Okay so S=1/3 returns ES=1.4 and a probability of finding 1+ cases (one or more) is simply 1-E-S (as (ES-1)/ES or total less none divided by total) or about 1/3 for practical purposes. For compact convenience denote this as P11 where first subscript denotes integer multiple of S=1/3 and second denotes incidence 1+ here. In general then with the simple rounding on powers of E and 2, P3S1=1-2-3S/2 and P3S2=1-(1+S)2-3S/2, the latter returning about 1/9 when 3S=1 and so discounted for practical purposes being less than the 1/6 suggested on basis everyday expectations in of dice-throwing. So getting one positive from 12K tests would indicate 500 fatalities is a plausible prospect but more should be expected if there was more than one positive test.
Okay so what about 3S=2? Well P21=1/2 and P22=1/6 which just touches our significance threshold so two positive tests would then imply up to 1000 fatalities is plausible. For 3S=3 we get P31=2/3 and P32=1/3, the latter meaning that 2+ definite positives could well mean up to 1500 fatalities but the sampling is now insensitive on this threshold and really demands discrimination at the next higher level of 3+ cases for which P3S3=1-(1+S+S2/2)2-3S/2 so P33=1/6 which again touches the threshold of commonsense significance . With 3S=4, P41= 3/4 and P42=1/2 and P43=1/4 so the sampling has again become insensitive and would need next higher order of 4+ cases for some assurance on 2000 fatalities. In fact none were recorded in the fairies thing so the implication of our reasoning here is that 2000 fatalities is likely on the high side, moreover dismissing the 2 possible positives would leave only one and that would put even 500 on the high side of threshold expectation.
The computer modellers surely suggested an uncertainty band for their figure and I could believe it was as much as factor two either side ~ if they did then the recently publicised thing added absolutely nothing new beyond reinforcement which is fine as an academic adventure but not a lot more. What’s been said here is about the best can be said but it’s still useful as an immensely less depressing portrait than the fairies’ painted in emphasising unlikely numbers well on the high side suggested here, in turn inevitably and probably intentionally initiating a feeding frenzy of “epidemic” by ill-informed so-called science specialists across the media.
Of course it’s possible that their long shot conjectured doom turns out to be true (probably not but nonetheless possible) but even then considering the exceedingly small numbers (in actuarial terms, no comfort to afflicted families) it really would make much more sense to spend scarce resource on individual support, even R&D as long shot for better palliatives (even therapies?), than burning bread to dance with pinhead fairies whilst playing deified dice. My commentary cost nothing apart from my time but that in any case was uncharged. I imagine that testing 100K samples with an exotic assay could cost upwards from £1M, maybe as much as £5M, and that would go a long way as assistance for the few hundred or so afflicted families identified to date.
PS I haven’t seen any of the sums but imagine it’s quite possible that essentially the same tricks have been pulled pretty much across the board in apocalyptic arenas ~ gemganic and climate come to mind as clear candidates where significance is pulled from super sparse sets, the former regarding pollen dispersion and pesticide contamination, the latter on storms and floods ~ thinking especially attribution assigned to the French heat wave last year which was not merely blamed for 20K untimely deaths but has since been assigned extraordinary significance from being asserted a 1 in 300K extremity on the basis of historical records. Do I really believe that such an enormous number denotes a defensible metric ~ not really, not without confirming it for myself! After all so far as I know no-one so far has adequately explained the much higher than expected incidence of thermal blips implied from ice-cored records ~ until that’s done why should we believe in much more than an associative correlation with anthropogenic atmospheric gases whatever causative connection may be clamoured by vested interests. I really don’t know what the truth is but I doubt it mirrors the mantra intoned for apocalyptic attention!
I have taken to calling these temptations dispositional contamination on a spectrum from ignorance to intent and it’s fiendishly difficult to disentangle this stuff in computational correlations, especially when nonlinear complexity causes bifurcations bursting out all over and even more when those that are realised are artefacts of the “closures” rather than actualities from fundamental physics. Stretching partiality from truth through bias toward desire is strictly fraud, at least spinning and we’ve all seen what a dangerous game that can be when played by posturing politicians and their para-politico fellow travellers playing pseudo-science for personal promotion especially prevalent in environmental arenas ~ things like comets, mega-tsunamis, floods, earthquakes, climate-cooking, nano-goo, etc… come to mind. All most probably crap of course, as significant than fifties flying saucers*, but then who really cares that campus core communities know it’s only a few crackpot cuckoos who promote this pulp-fiction aided and abetted by adverts from advancement agency apparatchiks and obsessive operators manning our manipulative media. Spiked statistics stretching spiked science should be seen as a seriously scandalous if not slanderous slant! Enough of my meldrewismic moaning (for now)
* Irony intended because of course these things have become paradigm exemplars for machination masking by populist politicos in pursuit of alternative agendas accessing advantage for the fortunate few, with common and crude contemporaries being bread baked in kickback cul-de-sacs signposted as renewable and natural and sustainable…!