Bad assumptions cause incorrect conclusions…

Hmm. In my last post I suggested that I had reached a conclusion about the CSS and div-related image indexing test. I might have done, but I think that I jumped there. 

The original motivation for the test was to work out why certain images were not being indexed. Two hypotheses presented themselves – some slightly sloppy nesting of divisions, and a clear CSS hack.

Sure enough, when the various relevant pages were indexed and cached in Google, I found what I was looking for – that some of the pages didn’t appear to show the images in the cache. This would also tend, I thought, to support the hypothesis that the pictures not appearing here would be excluded from the image search results – because Google, having “refused” to cache the images on the page, would surely “refuse” again to include them in the index.

A nice enough hypothesis – and having been pleased with myself for devising it, of course I wanted it to be true, so started looking for results that would confirm it.

Those pages showed up with no images, and I published my immediate conclusions. However, I was looking at the cached pages in Firefox and Safari. Today I took a look using another browser that I rarely use, Internet Explorer. Using this browser, the images were visible. Google hasn’t “refused” to cache them. The hypothesis seems much weakened. 

I’ll continue to track what happens to these pages and the images on them, and report back. However, an important lesson has been learned from the experiment in any case: do not allow your desire to be correct skew your interpretation of the results that are returned.

Tags: , , , ,

Leave a Reply

You must be logged in to post a comment.