I am not in the habit of using a camera in public bathrooms, but since I haven’t written any post in the CrazyStats category for a while I figured this was worth taking the risk of being arrested. Last weekend, I had the honor of using a urinal which “saves 88% more water than a one gallon urinal”. A completely meaningless statement that masquerades as a statistic (presumably they mean “uses 88% less water”). How much water does a one gallon urinal save? I know how much it consumes (one gallon) but how do you define how much it saves? Compared to what? To a standard one gallon model? Well, a one gallon doesn’t save anything compared to a one gallon, so the urinal I used (if it uses less than one gallon) actually saves infinitely more water than a one gallon urinal. If you are going to make meaningless claims, why stop at 88%?
Marketing claims based on meaningless statistics. It’s not just for Cloud Computing.
To the various conservation laws from physics (e.g. of energy and of momentum), one can add the law of conservation of hype. In the IT industry, as in others, there is only so much bandwidth for over-hyped concepts. Old ones have to move out of the limelight to make room for new ones, independently of their usefulness.
Here is this law, I think, illustrated in action. After running a Google Trends report on “web services”, “SOA”, “virtualization” and “cloud computing”, I downloaded the underlying data and added one line: the total search volume across all four terms. Here is the result:
The black line, “total”, is remarkably flat (if you ignore the annual Christmas-time drop). There is a surge in late 2007 for both WS and SOA that I can’t really link to anything (Microsoft first announced Oslo around that time, but I doubt this explains it). Other than this, there is a nice continuity that seems to graphicaly support the following narrative:
Web services were the hot thing in the beginning of the decade among people who sell and buy corporate IT systems. Then the cool kids decided that Web services were just an implementation technology but what matters is the underlying pattern. So “SOA” became the word to go after. Just ask Sys-con: exit “Web Services Journal”, hello “SOA World magazine”. Meanwhile “virtualization” has been slowly growing and suddenly came Cloud computing. These two are largely an orthogonal concern from the SOA/WS pair but it doesn’t matter. Since they interest the same people, the law of conservation of hype demands that room be made. So down goes SOA.
The bottom line (and the reason why I ran these queries on Google Trends to start with) is that I feel that application integration and architecture concerns have been pushed out of the limelight by Cloud computing, but that important work is still going on there (some definition work and a lot of implementation work). Work that in fact will become critical when Cloud computing grows out of its VM-centric adolescent phase. I plan to write more entries about this connection (between Cloud computing and application architecture) in the future.
[Side note: I also put this post in the crazyStats category because I understand that by carefully picking the terms you include you can show any trend you want for the “total”. My real point is not about proving “the law of conservaton of hype” (though I believe in it). Rather, it is captured in the previous paragraph.]
It’s been a while since I added an entry in the CrayStats category. On my drive back home tonight, I heard a gem, courtesy of both British and French public broadcast. So I guess it’s not just a problem at NPR.
I was listening to a podcast from a history program from France Inter (French public radio). It was about the eruption of Mount Vesuvius that destroyed Pompeii and Herculaneum in AD 79. They played a clip from a BBC program that explained that the lava coming down was “five times hotter than boiling water”, a figure that the French host later repeated.
Never mind the fact that, on the Kelvin scale, the same lava is only twice as hot as boiling water. More on the siliness of applying percentages to temperatures here.
Another off-topic entry to add to the CrazyStats category. Today’s NPR’s “All Things Considered” included a report called “States Fret at Easing of Border Security Plan” which talked about “Operation Jump Start”, so described:
“For about a year, National Guard troops have been rotating in and out of outposts along the [US-Mexico] border. Soldiers stayed visible under blue tents right on the border to deter illegal crossers while scanning the landscape, reporting anyone who did cross.”
It then goes on:
“The deterrent worked. The number of crossers apprehended by the Border Patrol since last October is down by about one-third, while drug seizures are up.”
The implication seems to be that would-be illegal immigrants were deterred by the presence of the troops and that drug traffickers were not deterred but were more often caught thanks to the help of the troops (who presumably either directly caught drug carriers or freed up Border Patrol resources to go after them). Success! But what if the result had been the exact opposite? More crossers apprehended and fewer drug seizures. Couldn’t that just as easily be interpreted to mean that the troops helped in catching more crossers while providing reinforcements that deterred drug traffickers? When opposite results can be interpreted to both mean success the test is suspicious.
After the 12% temperature rise, I recently ran into another creative use of percentages. Since I expect to run into many more of these (based on how many I’ve noticed in the past) and since they’re fun to point out I’ve created a new CrazyStats category.
This instance comes from a print advertisement for Samsung TVs, stating that their TVs with a 16:10 aspect ratio offer 30% more viewing surface than a 4:3 TV. Sorry, I don’t have a link but this advertisement (for computer monitors instead of TVs) repeats the “larger than 4:3 monitor” claim several times, albeit without quantifying it. This comparison makes no sense until you fix one dimension. And obviously it is to the advantage of the 16:10 screens to fix the height as being common between the two screens and then compare the surface (but even then, you only get a 20% advantage for the 16:10 compared to the 4:3, I don’t know how they came up with 30%). But if you fix the width as being the same then it’s the 4:3 that offers 20% more viewing surface…
Not that I don’t agree that 16:10 is a more useful aspect ratio (that’s what I bought for my monitor at home). But the “larger than 4:3” claim is meaningless. Next thing you know, people will start marketing 4:3 monitors as “16:12” to make them seem “bigger” than 16:10 monitors.
Here is another one to file in the “lies, damn lies and statistics” category: an article dated yesterday titled “Dutch bask in warmest autumn in three centuries” that starts with “The autumn of 2006 has been the warmest in the Netherlands for over 300 years, 12.5 percent hotter than the previous year which was already a record, meteorologists said.” We find out later where this 12.5% comes from: “The average temperature for the months leading up to November 17 was up to 13.5 degrees (56 degrees F), as compared to 12 degrees last year.” Except that such percentages don’t make much sense when applied to units that have an arbitrary zero. The same calculation using Fahrenheit degrees results in only a 5% temperature rise. Use the Kelvin scale and you’re down to a paltry 0.5% rise. Now imagine that a city has a 0.5C average one year and 1.5C average the next year. That’s a 200% increase in temperature! And if you live in a place that at some point gets a 0 degree temperature average, I would recommend moving out before the next year because you’re very likely to experience a terrifying *infinite* rise (or decrease) in temperature the following year!
This article comes from Agence France-Presse. And the French education system is criticized for putting too much emphasis on Mathematics…