h a l f b a k e r yExpensive, difficult, slightly dangerous, not particularly effective... I'm on a roll.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
All the most important things boil down to correlations. High
intakes of olive oil correlate with lower heart disease; use of
condoms correlates with lower rates of HIV; greater use of coal
correlates with increased erosion of masonry. Everything is a
correlation, until the underlying cause/effect
relationship is
established.
So.
MaxCo is launching the Database of Correlatable Information.
Users can upload any data they like, rather like YouTube, but in a
limited number of standard formats. For example, you might
have
records of the rainfall in Leeds for the last two decades. Someone
else might have data on the frequency of mobile phone ownership
across Africa. Someone else else might have counted the number
of
swallows nesting in a barn in Tashkent over the last decade.
Everyone knows something.
Each dataset has to contain a minimum of 30 values (to make it
worthwhile). Standard forms will be available for values
distributed
over time (eg, swallow nests per year), space (eg, mobile phone
density versus geographic location) or both (eg, lightning strikes
on
aircraft divided by location and year).
Users will, of course, be able to browse this atticful of data as
they
wish, using a variety of graphical representations. I can imagine
many happy hours looking at the annual cycle of Accident and
Emergency admissions in Stoke on Trent, or the region-by-region
breakdown of otter sightings.
This, however, is mere frippery.
The main purpose of the database is to unearth correlations.
Algorithms will regularly trawl the data, looking for all possible
correlations between datasets. Positive correlations (maybe
between rainfall and road accidents), negative correlations
(maybe
between fluorescent light-bulb sales and sunburn), or correlations
offset in time or space. Day and night, the algorithms will trawl
and
sift, reporting all significant correlations.
Users can then browse these correlations. Does anything correlate
with the number of coffee shops per square mile? Yes - ownership
of
mobile phones correlates positively, whilst sales of tractor tyres
correlates negatively. How about otter sightings? They correlate
with local rainfall positively by time, but negatively with sales of
hand-cream in space. And strangely, cheese exports from Menorca
show a significant (p<0.02) correlation with typhoons.
A million datasets, a trillion possible correlations....Many will be
spurious (like otters and hand-cream), but some (such as between
television sales and influenza, or cheese and typhoons) will reveal
new
connectednesses in this otherwise uncorrelated and confusing
universe.
Superstition in the pigeon
http://en.wikipedia...ition_in_the_pigeon
A bit of B.F. Skinner's work that this reminds me of. [Spacecoyote, Feb 10 2009]
The Journal of Spurious Correlations
http://www.jspurc.org/
[phoenix, Feb 10 2009]
House of Stairs
http://en.wikipedia...liam_Sleator_novel)
A novel about conditioning similar to Skinner's pigeons, but with humans [simonj, Feb 10 2009]
Helvetica Scenario
http://www.youtube....watch?v=aY7XH2ulTEU
Can be managed by homoeopathic Hanley - note the comment by amishrobots. [nineteenthly, Feb 11 2009]
[link]
|
| |
... So, the inevitable buns for this idea will clearly correlate to the global economic downturn. |
|
| |
Start posting bad ideas, yes? |
|
| |
Ben Elton's 'Blind Faith' did something like this in a branch of NatDat (National Database) called DegSep (Degrees of Separation).
To quote:
"Trafford's department had recently been astonished to discover that the DegSep computers were not linking preference in pre-cooked meals with parental star signs; hence the fact that an individual with at least one Taurean parent had a very slight statistical likelihood to eat lasagne more often than a person with two Sagittarian parents had lain completely hidden."
I am sure that you didn't know this, of course! |
|
| |
Aha - it seems that Phoenix and Gnomethang have found
some prior art... |
|
| |
As long as the algorithms can actually assign a stastical signifigance to the correlation, I think this would be useful. Obviously it would only work for things that are readily quantitized, and of course correlation is not causation, but it could provide interesting areas of research (and quite a few doctoral thesis regardless of research validity). |
|
| |
To be fair, you have taken things a bit further.
My first thought was the works of the Daily Mail:
'Drinking Wine Causes Cancer!'
'Drinking Wine Can Help Fight Cancer!'
'Can Drinking Wine affect your House Price?!'
Repeat Ad Nauseum..... |
|
| |
sounds like a fun website [+] |
|
| |
//As long as the algorithms can actually assign a stastical
signifigance to the correlation// Should be no problem, at
least in a one-size-fits-all way. There are standard
algorithms to handle correlations of quantitative data, and
they assign a probability of the apparent correlation being
due to chance alone. Hence, p=0.01 means that the data are
so correlated that there's only a 1/100 chance of that degree
of correlation being seen by chance alone. |
|
| |
Hey! I've been to Stoke on Trent ! |
|
| |
Finding correlations is like playing billiards. To have it count you need to call your shot before you make it. |
|
| |
Sounds rather like my entire life. |
|
| |
[Custardguts], doesn't this mean we should stop posting _good_ ideas? |
|
| |
//Hey! I've been to Stoke on Trent //
Counselling is available to help you get over it. |
|
| |
Excellent - and I think it's jolly fortunate you managed to post this idea before we all posted these annotations on it. Imagine if you'd posted it a day later! |
|
| |
I've been to Stoke on Trent too. Is there a correlation between annotators on here and going to Stoke on Trent? |
|
| |
//sales of hand-cream in space//
I always assumed that the astronauts got it for free. |
|
| |
//Is there a correlation between annotators on here and going to Stoke on Trent?//
It seems increasingly likely. Not only did I live in Stoke-on-Trent for a while but I was also an emergency admission on one occassion.
Post-hoc ergo propter hoc. |
|
| |
<someone had to say it>
//Cheese causes typhoons// No, just nasty wind sometimes.
</shtsi> |
|
| |
So [DrBob], is mere presence in Stoke On Trent itself therapeutic? I have often heard the opposite. Are homoeopathic doses of Stoke On Trent good for geodermic granititis?
On a slightly related note, did anyone know that the ghost of Tchaikovsky's star sign is the Croissant? |
|
| |
Well that depends upon which of the five towns you are in and at what precise time. For example, the currently worst time & place to be in Stoke, spiritually speaking, is at the Brittania Stadium at around about twenty to five on a Saturday afternoon.
I'd rely on Anti-Cobble cream* for the granititis though.
*Thank you for another credibility save Wikipaedia! |
|
| |
//correlations offset in time or space// So 25 years before and 1800 miles away from a spike in coffee shop openings in Stoke on Trent, there was a rapid decrease in otter numbers. I think that's a good reason to ban coffee shops. |
|
| |
I went by narrow boat to Stoke-on-Trent. |
|
| |
Yes, both those annos make a lot of sense. I tend to equate the Potteries with the Medway Towns, which makes Newcastle under Lyme into Gillingham. Maybe there are subtle distinctions between the precise actions of different Potteries, so that Stoke is more directly for granititis, whereas Hanley is good at preventing the Helvetica Scenario. Certainly the calcium content of the ceramics would seem to suggest that. |
|
| |