Wikiseek seems to be the hottest news of the day but honestly it is just a poor mashup of the Wikipedia database, a poor quality search engine, and some PPC adverts. Whilst Wikiseek has launched in a huge fanfare, and admittedly is still in beta, my personal opinion is that Wikipedia are taking the wrong direction.

Wikipedia Search – How it could be done

1. Take the Wikipedia dataset – yes I know it is subject to occasional spamming, and frequently there are editor problems over who is notable, or should be included, but at the end of the day the majority of sites linked to are of high quality.
2. Wikipedia API – at the end of the day, the best Wikipedia search will come about by natural selection. If Wikipedia want peope to be able to find the best search results, they should give developers access
3. Shared Revenue – Wikipedia could easily request shared income from any search to monetise the content to help with further development.
4. User Choice – allow a user to select which search they use with Wikipedia – List each mashup that is created and meets the Wikipedia guidelines, and let the users decide which to use.

Collaboration Brings Results

The recent collaboration with the OPML file created by Toprank Marketing, SEO search created by Alister Cameron, The Megite created for the SEO and SEM industry, and my eventual SEO Search Toolbar Button is a good example of how one simple dataset can in a short space of time.

A Wikipedia Equivalent?

• Give Access to Wikipedia Data
• Create an OPML File
• Import Wikipedia Data and linked sites to a Google CSE (Custom Search Engine)
• Share the revenue

A Google CSE based on Wikipedia would be highly useful. Wikiseek isn’t.

SearchMe

SearchMe is the company behind the new Wikiseek.

Through its patented technology and proprietary vertical search engines, Searchme delivers more meaningful and targeted search results to its users.

I would argue that currently the results are not more relevant than the results that can be obtained using Google CSE – maybe that will change in the future, but I don’t have the benefit of hind-sight.

Data Access – a Darwinian Approach

If Wikipedia gave complete access to their data through an API, just like Open Source encourages software development, access to data would improve search.
Developers would be able to create niche specific search engines, encorporating the Wikipedia dataset, but also supplementing the data with their own lists of trusted resources.

Junk Results Examples

Results using Wikipedia’s own search are actually better than these few examples

HTML
Website
Ground Mole
I don’t need it but… – seems they are blocking some searches – Wikipedia’s own search found this.

You have to do a very top level search to gain any reasonable results

Mole

The results for Long Tail are not so bad compared to a normal Google Search, but where did they get those “Relevant Adverts” from?

Wikisearch, Bloggers and SEO

How does this affect you?

My own opinion currently is it doesn’t. Produce quality good enough to be linked to from Wikipedia, and from other people.

If sometime in the future we all get access to the Wikipedia database through an API, then we might start seeing some creative use of the vast amount of data that has been collected.

You should also listen to Danny’s Daily SearchCast – he rips the quality of the search results apart, and also buries them on the placement of the top Wikipedia results. He did a search on the same blue pills as I did.
It looks like he must have performed the search before me, as I wasn’t getting any results, just an error message.

What are your views?

The initial feedback I have seen from many bloggers is “more of the same”.

Google Co-op is not the same as Google Adsense for Search in many ways, and the new service can be used / abused in ways that Adsense for Search could not.

1. There are some distinct differences in the code that you include on your website
2. There are additional collaboration features
3. The terms of service have a different feel, and are missing a few specific clauses
4. Sites can be included in the results, but also blocked

There are a number of projects I have been trying to launch for the last month that would use Adsense for Search and other Google services to provide a unique service.

I have had quite lengthy discussions with the Google Adsense team, but although they have allowed a number of sites to use Google Adsense for Search in a similar way to what I intend, they are reserving this only for test sites. What I intended to do was combine a few of the ways Adsense for Search was being used, with the way another Google service was being used, to create something unique. Currently I am not allowed to do it.

Because of the differences in code (point 1 above) for Google Co-op, one of the ways I wanted to use Google results would be impossible to implement.

The difference in the terms of service however might allow for a partial solution, and thus I have sent an enquiry to Google to see if this will be possible.
Most people would assume that if there are 2 Google products, and you want to have them interacting together to provide a better, unique service, that Google wouldn’t object. There are even situations where a visitor to your website is allowed to perform certain actions with a Google product, and if they then distribute the result from that interaction, your Google Adsense account could be in jeopardy.
I am being very vague in much of the above. I treat my discussions with Google as private, and obviously want to be able to implement my unique solution sometime in the future, whilst retaining a competitive edge.

If you are doing anything out of the ordinary with Google services, I strongly advise consulting with them first.

Of particular concern is anything that modifies or generates Google adsense code, and doesn’t connect directly with their API.

Google provide a way of modifying Adsense code “on the fly” with their API, but most solutions take a shortcut and modify the code directly with substitution. An example of this are the many plugins for sharing adsense revenue, where all you have to do is provide an adsense publisher ID to get your share. Any site that does some form of direct substitution has to be authorised by Google.

It should be noted that Google Co-op Custom Search is a beta project, but it is labelled “beta” and not “labs”. That means this is a core feature of Google, unlike Free Search, or even Google Reader.

Further Reading

Search Engine Round Table has some solid information on the signup process for monetizing your search with Google Co-op.

Neowin has a short piece, but the comments made me chuckle. Co-op certainly has a different meaning to most readers in the UK, and it is nothing to do with Google.

Techcrunch has a relatively short article on Google Co-op.