Thoughts on Freebase

I have been evaluating Freebase a fairly new site that is aiming to create an open database of everything. Think Wikipedia, but with user definable relational structure.

Here are my thoughts:

Weaknesses:

• Permissions too lapse.
  Not extendible for uses such as a social network, could not easily be used to store private data for a website.
• No notion of functions.
  Wouldn't it be really cool if users could submit functions as well as types, e.g. a function from post codes to latitude and longitude.
• Can Metaweb Technologies be trusted?
  This is a lot more complicated than Wikipedia for other organisations to replicate. If Metaweb goes bust or bad is all the data lost?
• Needs more data.
  It will be interesting to see at what rate the size of the database grows and what the typical method of inserting data will be. Many users will not go as far as learning the API but the current user interface is a bit too clunky for contributors. Refactoring using the Freebase site is currently more-or-less impossible, and a lot of refactoring should be expected for an openly writable database of everything in such an early stage!

Cool:

• Users can create HTML only data driven web applications.

Possible (Mis-)uses:

• To store low sensitivity, yet private, data (using cryptography).
  Think of it as a replacement for Amazon S3.
• A truly open social network.
  At some point soon Facebook, Myspace, etc. have to face the openness interrogation that awaits most successful technology companies.
• Free hosting with unlimited bandwidth.
  When the site is out of alpha anybody will be able to query the database.

Snap

I have added Snap Preview Anywhere to my blog. What a great idea! Whenever you hover over a link you get a thumbnail of the URL it leads to.

Amazon must be upset too because these are thumbnails of the exact page, not just a domain home page and they are captured very soon after being queued -- much better than Alexa.

I have some quibbles from a business point of view though: They appear to be marketing themselves as a search engine; that isn't going too happen there is way too much competition and their results aren't even taken from one of the top three (Snap adds a great new level to this cryptic linking technique doesn't it?) The behaviour is added by webmasters themselves by adding a JavaScript to their page, I think the decision to see these previews should be a user decision and not a content provider decision; this means the realm of browser add-ons.

UK Postcode Coverage Map

On a website that I have been making I had to provide a "Find my nearest" page, one of those things where people enter their postcode to find the nearest outlet of the company in question. An easy application of Pythagoras' Theorem using the full UK postcode database that I was provided with.

In a bored moment today though I wrote a short little program to go through the database and mark a dot on a blank canvas for the coordinates of each post code. As can be seen I got a very realistic outline of Britain. Although, I was a bit disturbed by the shape of Wales so I overlayed my data onto a satellite photo of the UK, there does appear to be some sort of a problem, is my database broken?

The satellite overlay does do a good job of confirming that the less postcode dense mainland areas very often correspond to mountainous areas.

Update: Thank you Mapperz for pointing out my mistake (Northern Ireland postcodes work off a different grid system). Seeing as it is not just a simple constant offset between the two grid systems and the company I was making the site for don't have any Northern Ireland outlets, I have just omitted the BT* postcodes from my dataset leaving me with this map.