The Server Room

Searching ‘05-style

In school it seems that kids nearly always go to google to search for anything. The majority of teachers I’ve come across see no problem in that – it’s what they do when researching. Increasingly, web-savvy people are turning away from the one-stop-shop approach to searching in favour of more customisable search methods like Rollyo (via mikeoliver.org). Rollyo is going the right way in that it lets you decide which sites you trust on particular topics, but it doesn’t address the problem of certain sites being poorly indexed by generalised search engines (Rollyo is powered by yahoo! search).

Image search is one example of a problem area. Google images is pretty good, but it’s far from flawless:

• As people have pointed out their results tend to be badly out of date, leading to dozens of dead links. I first noticed this when knocking together a web app to return google image results in date order (using the EXIF info in the thumbnail) – the most recent results were over 6 months old! I had been told that this was because spammers were cunningly using google to host images for their spam emails by letting google index them then linking to the preview thumbnails. Grr. (That explanation hasn’t been verified btw)
• Many sites don’t want their images to be indexed by google. Perhaps they have their own, better way of indexing the images. Sometimes the image site is owned by the competition. Either way, the results from some important sites don’t show up.
• You don’t know the copyright/licensing status of images unless you trawl through each of your results manually, one-by-one looking for the relevant information.

In this case, pupils and teachers would be far better served by casting their net wider than the reflex google image search. Having a good jumping off point in each area is important – in this case Wikipedia has an awesome page of Public Domain image resources. Mentioned on that page is flickr. Despite being bought by yahoo! recently, it’s still difficult to find fault with the site. Tucked away on flickr is a really useful page which lets you search for images with specific Creative Commons licenses. Although people often ignore the copyright on images when putting together intranets etc. it’s good to know that it’s easy to do it the right way.

A similar method can be applied in other areas – you wouldn’t google for the precise wording of a debate in Parliament, you’d use the Hansard search page. The same applies for many specialist subjects (medicine, technology etc.), or does it?