Why I brought back Tilting Pt2.

It seems Bengo has updated again and is still on about Google Trends. He still doesn't seem to know how it works, but that doesn't stop him returning to his old faithful of explanations.

People are lying about their stats and faking their analytics numbers!

He writes:

"I am puzzled that many major titles show big audience declines over the past two years, and wonder what is the cause. Confounding the mystery, I found a small study that looked at sites outside of webcomics. It found that data from Google Trends and data from the same sites' Google Analytics matched closely."

I fixed his link there, seeing as the one on his site was broken. One theory that Bengo hasn't mentioned is that the way people read webcomics is significantly different to the way that people visit other sorts of sites and so user behaviour varies greatly.

The other option is that he just doesn't know how to read a graph.
Go down to the bottom of the article and there are the example comparisons. What do you notice about them?

They don't all match up. As the writer says "In many cases last quarter, more recent period, although at Analytics stats are nor clearly going down they do at Trends". The typo's left in from the other site. What's the explanation for this? Well, they make a guess, but their final words in the matter, "who knows".

Google themselves have a possible explanation.

"15. My website's information in Trends for Websites doesn't match my own data. Why?

It's important to keep in mind that all results from Trends for Websites are estimated. Moreover, the data is updated periodically, so recent changes in traffic data may not be reflected. Finally, keep in mind that Trends for Websites is a Google Labs product, so it's still in its early stages of development and may therefore contain some inaccuracies. In the future, we may consider ways to improve the accuracy of Trends for Websites by allowing owners to contribute additional data from their sites."

That bit I bolded is important. It effectively means one thing. Trends is still in BETA and as such is still going to be full of bugs. Combine this with the differences in the way webcomic sites are used and you might see why it will have some odd results.

It also would be nice if Bengo could give us a link to some sort of evidence of the claim that "click fraud has risen dramatically every year and passed 25% a while back". Quoting statistics without sources is pretty useless and very sloppy journalism.

Another thing mentioned in the lasted entry on The Floating Lightbulb is that "there was a record amount of kook mail, all from the same Google account but using different names, genders and nationalities."

Well, we have some of that kook mail here. It seems one of Bengo's disgruntled readers has found this blog and taken me up on my offer to post any comments that Bengo wouldn't.

I'll preface this with the note that while the author was accused of using multiple personalities she stated in the email to me that she hadn't in fact sent a single email and had only attempted to comment under her own name. Here's the post that started it all:

"Hello again Bengo,

First off, may I congratulate you on becoming a running joke upon the webcomic list (both site and chat room) before asking you why you continue to spew unsubstantiated (and may I remark, poorly researched) garbage, you have not only show that you can not grasp the fundamentals of percentages, but also that you are a most paranoid, crackpot, conspiracy theorist and must be treated as such.

Congratulations must also be given as the above makes you a worlds first, the worlds first webcomic crackpot conspiracy theorist, how, may I ask, how do you feel?

You are not a researcher, as a researchers duty is to check and double check facts before sharing her/his findings, you admit that you do not, on this point I wish to share something with you, the first rule of research is that you eliminate any possible fallacious sources or information, it does not seem that you check the factual content of any information that you acquire from sources that might be lacking in integrity or perhaps showing bias.

At school, one should have learned that statistics are hardly a definitive source of information and thus should be views with a level of suspicion as should open sources like wikipedia that is so often proven wrong. May I also remark upon your most juvenile actions, by this I mean disabling your commenting to cease chatter from others who find flaws in your rants, also I must ask why do you continue to batter the halfpixel guys? Is it perhaps that it didn't work for you, if so, have you considered that comics, like art in general, is a very variable media and what worked for someone else, is hardly a guarantee of success as art (of all kinds) is hardly a science.

I wish to enquire about this store of yours, may I ask what was the url to this, perhaps some information would give credence to this claim, as on its own, hardly holds water.

Please get back to us,

Jessica Ottowell,

Creative Genius and Head of Development

fiveCARDSdown Group of Companies"

P.S.
Is Bengo including himself when he mentions "egotists, who commence axe-grinding at the first sign of dissent, and are not gracious to critics"?

Why I brought back Tilting Pt1.

First to make something clear. I didn't create Tilting at Lightbulbs. I only brought it back.

You see, the first Tilting at Lightbulbs was a blog dissecting Bengo's own blog, The Floating Lightbulb, pointing out the flaws in his theories. Unfortunately, I only discovered the blog through some cached pages after it's demise.

If you're interested (and why wouldn't you be if you'd bother read my version of this blog?) here are the pages:

Page one, page two, page three, page four, page five and page six.

The Solution has some interesting things to say about our friend Bengo and his articles and I was disappointed that it wasn't still around, even more so considering the change in commenting that The Floating Lightbulb has had recently, so I decided to bring it back.

See, I find it slightly worrying that there are a number of people who don't think that the Lightbulb is effectively the rantings of a biased, jaded individual who does not do any real research.

An interesting recent example of this is his article on declining webcomic readership. In his article, using information from Google Trends, he makes a list of webcomics that have been seeming to lose readers, ones that have been going steady and ones that are in fact gaining readership.

It seems like it could be a good base for starting research, but as he says he doesn't "know where Google gets their data". Okay, it's statistics based on unknown sources, so he looked into other places to see if the statistics matched before posting his article, right?

Wrong.
It's true that he admits he doesn't know the factual accuracy of the data, but he still writes a nice long article based on it, an article that would have been much improved if he'd made some serious attempts to check these statistics against other sites, for example Project Wonderful, who relay public statistics for the sites that chose to host adverts with them. That would have been a start.

Well, no double checking of statistics, but did he check the sites in question to find out what could have caused it. Apparently not, as the first person to comment, a user called amrothery, on the article other than himself pointed out some interesting ideas. I'll repost them here:

"Of course, some strips might have explanations that are even simpler. Dr. McNinja's drop seems to coincide with the switch to full-color updates. Diesel Sweeties looks to start dropping around the time Stevens left newspapers. Achewood's numbers suggest that people who came for the heavily-hyped (and highly praised, at least among a couple forums I frequent) Great Outdoor Fight saw no need to stick around once it was over. This is pure speculation on my part, but Girl Genius's declining numbers may be a result of the update schedule effectively getting cut in half when the last of the old print pages were posted."

So, some of the comics that have shown a decline according to Bengo's Google Trends findings seem to have good reasons to. A comic halving it's output should definitely affect the amount of traffic it receives for example. Now, I'd be interested to know how long it took amrothery to find this out and why it was beyond Bengo, who has in the past referrred to himself as a journalist, to look for this sort of information.

How did Bengo respond to this information? Well, he did respond to the post, but in true Bengo fashion he managed to skillfully avoid any mention of the of the examples above.

He did however have this to say:

"I expect Google gets its data from the same sources the other big analytics companies (like the one Adobe just bought) get it: from ISPs, with help from toolbars and statistical devices. The difference is Google has the might to get more for its money, and probably buys deeper data, or even gets it in exchange for services. But even Google is stingy about buying and storing data for sites that no one will ever check."

He expects they do.
Great journalism there, just great.