DRAFT – NOT FINAL COPY – check over the next few days for updates.
The Climate Reality Project’s recent 24 Hours of Reality: The Dirty Weather Report has produced some interesting numbers regarding the view counts of their 24 hour event. The event was on the 14th and 15th of November 2012. Currently, 7 days later, according to a Google search of the sites climaterealityproject.org and climatereality.com for the term “million” within the last month, they do not seem to say anything more specific than it was “watched by millions”.
Gore, et al, claim their Climate Reality Project’s 24 Hours of Dirty WX stream attracted 16M views. There is a supporting image, full of boast, floating around on the web that compares “final” view counts of the Dirty WX stream with the WUWT-TV stream with the in a side-by-side format:
That image is linked from, among other places, Dana Nuccitelli’s comment on
I know it’s only a throw-away thing, but it’s good to occasionally fact check inconsequential trivia. So here goes. It also usefully clears a minor doubt of mine that the screenshot collection I have was somehow “off” with respect to the view counts relative to the video and the time. The “Dirty WX” image in the comparison is in the time between two of my screenshots (taken 30 seconds apart), where the view count changes from 15.9M to 16M, so I don’t actually have that exact screen shot.
Just to confirm that the image in the comparison is genuine it is possible to find my two images either side of the image from the comparison in the video currently available at
My “before” image, from time 00:59:38 (showing 15.9M), which can be located in the video by dragging the little circle thing on the horizontal bar of the video player to about 36:26 (that’s not very accurate I’m afraid). Click the images to embiggen.
My “after” image, from time 01:00:09 (showing 16M) is on the video at about 36:55.
The screenshot from the side-by-side image is at about 36:47, which would be within seconds of 01:00:00 on my system.
I actually have a close-enough screenshot for the WUWT-TV feed for that time (the “7″ is just scrolling into view).
So far so good.
I am disappointed that the comparison didn’t go on to give the final count. Let’s see the screen shots for those.
(sarc)200,000 views for the end titles and black screen before the stream was turned off is pretty darned good. Well done. (/sarc)
The WUWT-TV continued for another hour after that time. It finished with … see for yourself.
Admittedly, 16,612 views isn’t as impressive as 16.3 million, but at least there were real humans behind the numbers.
Some plots from the period when the “current viewers” count was available in the climaterealityproject’s Gore stream, before the display of that value was turned off for some reason.
This free WordPress doesn’t seem to let me upload general files directly to this blog, so I have uploaded to dropbox here:
These are all the viewing figures I have relating to the Gore-a-thon stamina contest. The early figures include current viewer counts and “total views” for the Gore event, and the WUWT-TV streams. Fairly early into the 24 hours the stream of Gore stopped providing current viewer counts. The WUWT-TV counts have only been transcribed up until that change (and from the screen shots seem reasonable after then anyway).
Just to get something “out there”, here is a graph. It is derived from view counts from the Gore-a-thon, manually transcribed from the
core sample screen shots of the view count information from a Ustream video player window.
The data I’ve collected might should be taken with a huge pinch of salt. It is entirely possible that anything unexpected is due to the way that Ustream.tv collect or display the information or the information may be delayed due to cats on the internets. Apparent gaps in the data could be Ustream somehow failing to count at that time.
The first plot is of calculated “views per second“, of the last few hours of the webinar (which finished at about 01:00 UTC).
At this point in time, the view count is only available in 100,000 (0.1M) unit increments, and the reading is sampled (as a screen shot of the video) about every 31 seconds. The “steppiness” is basically due to the 31 second sampling and the 0.1M count increments.
That plot was a bit busy. Let’s make it a little more visually appealing by graphing the average of two adjacent count values (I expect it’s not entirely statistically valid due to the varying time between samples, but it looks pretty).
The + marks (only in the “smoother” plot for now) indicate the possible per-second counts due to the granularity of the count values and the sampling period, the averaging allowing values about mid way between the +s. The plot at this resolution would be fairly smooth below about 175 views per second (or 175*60 = 10,500 views per minute).
700 views-per-second is 42,000 views per minute, or 2.5 million per hour.
The whole thing certainly requires more explanation. More later (possibly in a couple of days), including the raw transcribed data and explanations of how the data is collected. The original screen shots also include the WUWT-TV stream, so the viewer data for that could also potentially be extracted beyond the hour or so of data I have already transcribed.
This is a graph of the reported “views” v. UTC time of day.
Note that the “event” started at 01:00 UTC, and I didn’t start to collect data for the first hour so the first data point is at about 02:13 UTC. The gaps in the left half are due to me either not collecting data, or the data not being available on the video. I appear to have full coverage of data on the right half, so gaps are due to the count not updating.