Tuesday, June 30, 2009

Wither Twitter?

After distributed computing efforts such as SETI at home, and Climate Project came GLOBE, a world wide effort to involve students in gathering ground truth for environmental and climate studies and Weather Bug, a distributed set of weather stations available to classes, media, and on your Blackberry. Motivated by early efforts to gather climate and weather data over new media Eli asks, wither Twitter and is not very optimistic. From the Urban Dictionary

freep: To slew or cheat an online poll by repeatedly voting (clearing cookies, using proxies) or to make a blog appear to be commented by numerous posters by the same means. (From the practices of the Free Republic or "freepers")
The results of the CNN question of the day were running 70:30 in favor until an hour ago when it got freeped.
Alas, without enforced automatic quality control or restricting the contributors to trusted sources, something like this is doomed which is too bad, because involving the public in climate and weather sciences would be a valuable thing. Not only would coverage increase, but feedback could be used to provide a valuable introduction to science, mathematical and statistical concepts and ways of posing questions and answering them.

Comments and suggestions?

11 comments:

CapitalClimate said...

After having been actively involved with Internet development since the earliest days of the web, the Climate Capitalist has reluctantly concluded that the www is the implementation of Gresham's Law at light speed.

CapitalClimate said...

There was a discussion of this concept a few months ago at Climate Progress ("So you want to be a citizen scientist"), in which my comment that the concept of "citizen scientist" was extremely naive was greeted with, shall we say, a great deal of naivete.

Horatio Algeranon said...

It might be possible to make it work.

After all, it works quite well in the case of variable star observation.

In fact, some amateurs are quite "professional" in the way they approach science. Some (eg Wayne Lowder) even rival the best professionals and their fancy equipment.

The cases are actually similar in one all-important regard. In both the case of variable star observation (actually done by eye) and the case of weather monitoring, large numbers of measurements are averaged together, reducing the overall error.

If enough measurements are taken, it can actually more than make up for the lack of precision and accuracy associated with individual measurements.

The latter point is something that many people -- even a few who call themselves meteorologists (Horatio will name no names) do not seem to appreciate.

John Mashey said...

I think the Rabett doth protest too much.

Freeping works because:

a) The overhead of doing it is very low, and can be done by indivduals working by themselves.

b) And it is hard to provably detect.

Both of which show why most online polls are silly.

However:

a) If you have to have a weather station, that is a barrier to casual freeping.

OR
b) When you report data attached to geographic location and date/time, one can surely analyze the significance, find outliers, etc. It would take a serious coordinated effort to truly skew things. Somebody might do it, but it seems like harder work than the online poll gimmicking.

Personally, I think the value of motivating students to gather and analyze real-world data is worth a lot... even if it doesn't generate a lot of real data for science.

and this is not just TwitterScience :-) I think we should reserve that for the electronic equivalent of Gish Gallop.

amoeba said...

Amoeba

Very slightly O.T.
If I remember correctly, the Weblog Awards allowed repeated voting and as a result, the Best Science Blog award was awarded to the execrable WUWT.

Any online vote that doesn't as an absolute minimum, log IP addresses [or take other precautions] to filter multiple votes is thereby rendered pointless; vulnerable to automated vote-rigging and concerted attacks.

Anonymous said...

The difference between astronomy and climate science is that there are huge political implications in the results of climate change theories. Astronomy has no impact at all on our daily lives. So there is a huge incentive to distort climate science according to one's political inclinations (unless one is a pure as the driven snow kind of scientist, of course...)

-WilliamT

Horatio Algeranon said...

William T says "The difference between astronomy and climate science is that there are huge political implications in the results of climate change theories ...[and] a huge incentive to distort climate science according to one's political inclinations "

Not sure how "huge" the incentive is for the average person. Personally, Horatio thinks most people -- especially kids-- would probably be very honest about the whole thing,

Having said that, there are some who would try to "game" the system, no doubt about it.

But as John points out above, there are ways of spotting and weeding out efforts to do so.

"Outliers" are really not that hard to spot, especially when you have lots of coverage (ie, lots of weather stations all over, including traditional weather stations run by actual scientists)

GISS already compares temperatures at "neighboring" weather stations to spot potential outliers and find and correct for systematic biases (due to the urban heat island effect, for example)

And if the errors introduced by those trying to "game" the system were purposely made small to evade detection, their impact on the overall result would be even smaller due to the averaging process (unless everyone is doing it and in the same direction, of course)

There is no "guarantee" that such a network of "citizen scientists" would not be "undermined" by dishonest people in the case of weather monitoring, but it might be worth a try for educational purposes if nothing more, as John pointed out.

bigcitylib said...

Twitter feeds are pathetically easy to hijack. Thre is usually a short number/letter combo code naming the feed, and anything with that code winds up in the feed.

SO for example, some very nasty anti-union sentiment has been posted to the CUPE feed up here in Toronto during our ongoing garbage strike.

Paul said...

Please look at the Citizens Weather Observer Program among others. CWOP is receiving data from over 8000 observers. They run a quality assurance check invloving the comparison of near by stations.

Many of the stations are of near professional quality and many observers check their digital results against good quality analog instruments. The CWOP mailing list strongly discourages climate politics. The discussions are obsessively technical and data quality related.

http://www.wxqa.com/

Paul Middents
CWOP station DW1622
Weather Underground Station KWACOPAL3

http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=KWACOPAL3

Horatio Algeranon said...

Thanks Paul.

I think the "pride of ownership" issue plays a big role in making such things work.

When people become intimately involved with something like that, it becomes their "baby", which means they become very careful and serious about it -- and protective of it, on the lookout for anything that could mess up the results (especially purposefully).

PS Keep up the great work!

Anonymous said...

Thanks Eli for bringing this up.

After someone pointed out how silly it was for Anthony's friends to be taking all these pictures of weather stations WITHOUT bothering to take temperature measurements, I started thinking just how hard could it be to deploy a few thousand temperature sensors.

The kit would have some sort of sun/wind shield, sensor, and web box. 100$ ? Batteries not included.

Calibration? Keep a couple dozen NIST traceable thermometers moving around. The station keeper clips the thermometer next to the sensor, a little camera takes pictures of the (serial numbered) thermometer for a day. Upload picture. Repack thermo, off to next station.

I am sure there must be simpler ways to do this, but this would work.

Sea Ice Outlook, high stakes:

http://www.arcus.org/search/seaiceoutlook/index.php

Yuebing

(btw if you google Rabett, a picture of the Cadillac comes up)