cappucinoLike I was saying to Joe earlier, every time I speak to people about this restaurant website they seem to want to get involved or they have some excellent idea to add to the concept.

So the basic premise is a web 2.0 (and by web 2.0 I mean user generated content etc) restaurant reviews website where the reviewers are broken up into various bands according to their karma and the number of reviews they’ve written. New reviewers have their first 5 reviews moderated. Once they’ve got past their “New Reviewer” status they become “Reviewers” who can invite additional people to join the site but they only get 1 invite per review they write.

All the reviews are rated in digg’esqe manner and reviewers gain or lose karma (trust) accordingly.

Reviewers who achieve certain levels of Karma and write a minimum number of reviews become “Food Gurus” who have additional functionally available to them.

Food Gurus are the people who

  • Moderate the “New Reviewers” reviews for quality etc
  • Are also responsible for updating the non-review based restaurant data (addresses etc)
  • Respond to reviews that have been tagged as “inappropriate” or “suspicious” and potentially banning dodgy reviewers etc.

The idea is that the trust based invite system will hopefully keep the site as honest as possible.

Also, I imagine there might be a point in getting the actually restaurants to have a representative user on the site who can respond to praise and criticism and alert Food Guru’s of any errata or dodgyness on their restaurant’s profile.

Since reviewers have karma it makes sense to have an algorithm that determines the restaurant’s average rating based on various factors including karma etc. I like the idea of the age old 5 star rating but how we get to that number can be quite interesting. Firstly we’d obviously bias the result in the direction of the reviewers with higher karma. We can also positively bias the most recent reviews so that the result is a recent reflection, potentially setting a cut off date so that really old reviews are ignored outright.

One of the more interesting debates is the question of what is a “review”. In my original idea I kept is uber simple: 160 characters and a star rating. Is this too little? What is too much? Keeping thing simple makes sure that the reviews are actually readable; allowing a site user to actually read a number of reviews in a short amount of time. But since we’re building the system why not take the opportunity to build something that can accommodate all kinds of data, right down to how clean the bathrooms were. I think the answer lies somewhere in the middle, with the minimum 200 character review and a star rating as the base of the review but the site allows you to fluff up the data as much as you like.

Because no system would be complete without an acronym, I hereby introduce URT or Uniform Restaurant Tag, a unique tag for every restaurant in the entire world. We’ll have to think this one through carefully so that we don’t get caught by our own short sightedness when the site begins to take over the planet but you get the idea. This tag will be visible on each restaurant page so that it’s easy to copy and paste. The tags will be used to tag blog entries, flickr sets, technorati etc. This data can then be pulled in or linked from the restaurant profiles.

Joe’s been down the road before and one of his ideas was to come up with a XML restaurant review markup language (RRML). The basic idea here (I think) is the ability move the data around and store it easily. I’m not convinced of the storage thing yet but I definitely think that coming up with a markup language will ultimately aid the sites expansion as various 3rd parties use the data. (and link back to it)

nbm suggested setting up trackback functionality so that we know who’s linking to what and potentially push people to relevant blogs.

One of the most cringeworthy but potentially clever ideas is to get pro reviewers from traditional media (newspapers, magazines etc) to take part. This could help grow the site really quickly if the site gets adequate exposure.

We can also build in the ability to have Top 5 lists. Something like Top 5 Burgers in the CBD or Top 5 Sushi restaurants in Cape Town.

Lastly, but not leastly; Eatups: Meetups where you eat. We could potentially have a lot of fun organising big dinners at various restaurants with like-minded food people and have an en-masse review afterwards. Discounts?

I’m sure there are one or two ideas that I’ve forgotten to put down but I feel better now that most of it is committed to blog. (I must automate my backups)

Please feel free to give me ideas or criticism. Like I said the time before, I’d rather this gets shot down in flames as a bad idea before I stupidly spend any real time on it.

ath,

j.