Why ads?

Frequently asked questions

Why does it take so long to compute some stuff?

The main reason is that last.fm limits the number of requests to their services to 5 requests per second. While this is perfectly understandable it also means that this number of requests has to be shared among all users that are using the Tools. Sometimes the computation of a single method like the Soulmates needs heaps of calls in order to be completed.

An example how long it would take in worst case to compute the soulmates for a user with 50 friends:

  • Fetching friends: 1 request
  • Fetching taste signature for user: 2 x 50 = 100 requests
  • Fetching taste signatures for other users: 50 x 100 = 500 requests
  • 601 request / <=5 (requests/second) = minimum of 120 seconds

However I am using a sophisticated caching strategy to overcome this limit. Nevertheless it might still take some time, especially if a user has a lot of friends, so you won't get around being patient :)

I have made new friends (or other stuff changed) but it is not reflected in your data?

This is mainly because I cache the data to reduce bandwidth and load on my and last.fm's servers. Therefore the data is not perfectly up to date. It might take up to 4 weeks until some changes are reflected.

Your soulmates scoring is so different from last.fm's tasteometer?

Example distribution Very much so. I am using an own algorithm to compute the similarity between two users. The main principle is simple: I take the toptags and topartists of the two users and do a simple weighted vector comparison. (If you need to know more about it drop me a line). This does lead to the desired result: A user compared with itself will give a perfect match while there will be no match at all if the taste is completely different.

The main consideration was to compare both, the preference in particular artists but also in musical styles. Typical results will however not group around a mean value of 50% but rather around 15%-20%. Tastes are just that different :) This is why I also use color coding to show that low values don't have to be that bad. (For the statistically interested: It does resemble a Gaussian distribution though).

The other difference is that my comparison is only based on a twelve month period instead of the overall comparsion in order to reflect changes in musical preference more quickly.

The design is broken, will you fix it?

I am only supporting the popular browsers (Firefox, Safari, Opera, Chrome, Internet Explorer) in their most recent versions. Please report design errors, but only if they have occured within one of those browsers and you are sure they are pretty much up to date.

If I embed an image, do I have to update it manually?

No. The images are updated automatically once in a while.

I like what you are doing, how can I support you?

Cool, thanks. First of all, spread the word and give feeback. And if you really really like it you can prolong my Last.fm subscription. :)

Yeah I know, but I couldn't avoid it anymore. This site has become quite popular, which I appreciate a lot, but as it is running on my own server it also produces costs regarding bandwidth and server rent. Try to see the positive side-effects: Lowering my expenses enables me to further extend the site.