Docur blog

Site Improvements April 2018

Hi, Here are some site improvements that have recently gone live and will do soon:


Lists are back! – new and improved with much better usability and the ability to make private.


Massive improvement to search functionality. New engine that provides an indexing and fuzzy search capabilities.


Improved matching on related docs (still not completely perfect though).


Sever response speed should be better.


To make the first two points above more useful the site will soon start connecting to a number of external databases. This will give Docur access to around 50,000 documentary trailers and other information. All of the backend stuff has been done, but the question of how to implement this appropriately while keeping the same feeling that the site already has is a tricky one. It will probably be the case that docs which only have a trailer will be demoted, in that they will only be viewable if the user asks to include trailers only (they will also be clearly marked as trailer site-wide.


The aforementioned function raises the question of how many of those documentaries which can be crawled have been made freely available to view online on reputable streaming services by the filmmakers. It’s safe to say that while writing a program to collect accurate information about a documentary can yield highly accurate results, writing one which can determine the exact video seems futile. For there is a lot of BS on the web. But it is a cool idea and maybe a works projects for the site and not something that would have direct write access to the database for a long time.


While problems like this seem deeply impossible to do without an extreme amount of horsepower there may be a way of lateral thinking the problem. Despite this pessimism it definitely can be done. Google rank videos very well and they blatantly doing this with a comparison function which ultimately weighs very significantly against other factors in the algo until they are called out upon it.


When the above said is implemented it may be viable to make available some cool search features since it is now an index.