The Floating Lightbulb updated again and Bengo is still talking about Google Trends, but at least now he's doing some research, or in part having research done for him. Maybe it would have been a good idea to do some research first?
Bengo has had a go at comparing data from Alexa and Google Trends, with the intention of looking for similarities. What do the results show? Well, they don't match up very well. This really brings the validity of the original Google results into question. One of them is definitely wrong. Maybe both of them are.
Bengo managed to misread one of the graphs in his own blog, resulting in a false positive when it came to matching results from the two services. His response when it was pointed out:
"No, you're right. I overlooked the fact that the spike occurs in different years on each graph. I figured the data prior to the spike didn't meet the GT threshold, and that seemed a reason why it might not appear.
That appears to weaken the correlation with Alexa significantly, by replacing a strong match with a weak one. (Personaly, I would consider it a complete miss - GB)
I think, in the end, we will probably find that Alexa isn't much help here, though it was interesting to explore. Use of Alexa raises questions about data gathering methods that fuzz up whatever data you do collect. A solid answer should not have to depend on Alexa."
Wait, what? Is he suggesting that Alexa is at fault here? Does he have anything to support this suggestion? Anything? Because he sure as hell hasn't offered anything up here. This is classic Bengo. He has decided already what the correct answer is and when a source contradicts his 'findings' he seems to decide that that source is the less reliable.
If one is going to be further out than the other then I would think the one still in beta is the most likely. I don't know that it is, but it seems the most likely.
Now, amrothery has supplied an interesting chart that would seem to support a link between Alexa and Trends when it comes to webcomics, so maybe the discrepancy in Google Trends is in fact with sites other than webcomics. It's a possibility.
I would like to read what amrothery's take on the findings is. In fact I would be very much interested in reading a blog by amrothery, because in the space of a couple of posts amrothery has appeared to be a much better researcher than Bengo has throughout the life of The Floating Lightbulb to date.
Edit: Here's a link to the permanent address for the blog post on Evil Inc. that Bengo mentioned.
Subscribe to:
Post Comments (Atom)
Bengo started with the assumption that Google Trends is correct, and when he compared it with Alexa and there wasn't good correlation, that Alexa must be wrong. It's just ambiguous data vs. ambiguous data - what's the point?
ReplyDeleteAlso notice how in the previous post he took a moment to again list his reasons for believing all web cartoonists are liars, when it has no place in the discussion. It's like saying "I don't know whether or not the defendant is a rapist. We can't say." during a trial about a parking ticket.
Gordon, Scott Kurtz here.
ReplyDeleteI posted this comment at Bengo's site. Who knows if he'll allow it. Meanwhile, I offer it to you for posting here.
------
Ben,
When Adsdaq purged all those webcomic sites from their system last year they gave a very clear reason for their actions. It wasn't click fraud.
When word got out that Adsdaq was performing so well for some of us, it caused a run on the service by a lot of webcomic sites.
Adsdaq told us that they dropped these webcomic sites because they were too image-heavy and didn't have enough text content on them. Mostly it was webcomics without frequently updated blog content. Adsdaq can't place context based ads on image heavy sites. Images don't offer context their system can use. Text does. So the sites were dropped.
Sites like PvP, which has a context rich blog that's frequently updated were not dropped as they continued to perform very well for Adsdaq.
This information could have been gathered VERY easily had you simply emailed any of the cartoonists that you're clearly NOT accusing of click fraud.
Way to do your homework, son.
TS, yeah, I think you've got it pretty spot on there. He made up his mind and then went looking for evidence afterwards.
ReplyDeleteScott, thanks for posting.
It looks like Bengo put your post up on the Floating Lightbulb after all. I wonder if the fact you posted here affected that decision. I know that Bengo's aware of this blog's existence.
amrothery here. I've also been seen in Bengo's comments (when he allows them) as Andrew-TLA (a reference to my own little comic, TRU-Life Adventures), and, rarely, simply Andrew. You might recognize me as the guy who ferreted out the ownership of webcomics.com, established that the xkcd forum message was obviously spam, and actually thought to check for updates on the Mike Peters lawsuit situation. Sorry for the delay in getting here, but I only just found it.
ReplyDeleteBengo's tendency to go off half-cocked and focus on a single dark theory when there are likely several benign factors in play amuses me.
Google is in the business of gathering information. It's in their best interest to convince people to use their tools, and so I put that chart together to see if perhaps they were using Trends to reward sites that use Analytics or AdSense, or punish sites that use other services. I tossed in the RSS feed column just because other people insisted on bringing it up as a possible factor.
The presence of an RSS feed has absolutely no bearing on a site's Trends status. I'm willing to say that with all confidence. I'm also mostly convinced that Google's not actively manipulating Trends data in their own favor.
I take no responsibility for the Alexa columns. I gave Bengo access to the chart (which is just a copy--the original sits on my hard drive) to see what he'd do with it.
Until Google actually comes out and explains how they calculate their Trends data (and actually let us see Trends data for Google-owned sites), I classify it as a lot less useful than alexa or compete.com. They, at least, explain where their numbers are coming from, and an interested party can take that into account.
While I'm flattered you think I've got the makings of a good investigative blogger, that's not really where my interests lie. Sure, I'll fact-check someone into the wall if their assertion amuses me, but I'd really rather focus my energy on making and talking about webcomics.