Investigate if trakt.tv or Rotten Tomatoes API is better than themoviedb [experiment]

[ovidiuch]

• Must have JSONP support
• See if you can fetch all Movie/Person data in one request
• See how good "also see.." movies are
• See how good "know for" movies for person are
• Check out and acknowledge rate limits
• Compare response times (do N requests every rate limit + small rand for a minute or two and compare averages)

[ovidiuch]

Must have JSONP support

• trakt.tv does: http://api.trakt.tv/movie/related.json/368b0395732268808fa8d151d2484e8c/tt0468569/related?callback=paulinho
• so does RT: http://api.rottentomatoes.com/api/public/v1.0/movies/770672122.json?apikey=q48uhfxwd8a75n4xcykmym6k&callback=ciprian

[ovidiuch]

See if you can fetch all Movie/Person data in one request

TheMovieDB has something called append_to_response, which can be used to merge to API requests together, like movie summary + related movies. track.tv doesn't seem to, but the API responses seem pretty cool (cooler.) For example you can use imdb ids directly to target movies.

• Main movie entry point http://trakt.tv/api-docs/movie-summary - Contains needed meta fields, images, ratings and cast/crew (just not related)
• Related movies - http://trakt.tv/api-docs/movie-related - So we'd need two requests per movie... let's see if the content is better still

Rotten Tomatoes still can't aggregate more API calls in one, but they have a pretty cool movie summary and links to related API methods (which encourages hyper media and goes well with the principles of Cosmos)

• Main movie entry point http://developer.rottentomatoes.com/docs/json/v10/Movie_Info - Contains needed meta fields, images, ratings and abridged cast and directors (no writers tho, but we can go without, sorry writters...)
• Related movies http://developer.rottentomatoes.com/docs/json/v10/Movie_Similar- Still, two requests per movie... let's see if the content is better still

[ovidiuch]

See how good "also see.." movies are

First 10 related movies for "The Dark Knight (2008)":

Themoviedb:

• Batman: Under the Red Hood (2010)
• The Dark Knight Rises (2012)
• Batman Begins (2005)
• Justice League: The New Frontier (2008)
• Batman (1989)
• Reasonable Doubt (2014)
• Batman Returns (1992)
• Batman Forever (1995)
• Batman & Robin (1997)
• This Woman Is Dangerous (1952)

trackt:

• Scarface (1983)
• The Dark Knight Rises (2012)
• Heat (1995)
• Sherlock Holmes (2009)
• Ronin (1998)
• Hackers (1995)
• Training Day (2001)
• Drive (2011)
• In Brudges (2008)
• The Town (2010)

There are many things to consider here, and it's nice that themoviedb (I've seen examples with way more old movies show up in the related list for a contemporary movie), but the ones from trakt seem more varied and feel less like they were matched by a soulless computer algorythm..

RT: http://api.rottentomatoes.com/api/public/v1.0/movies/769959054/similar.json?apikey=q48uhfxwd8a75n4xcykmym6k

Weirdly, RT seems to only return 5 max similar movies.

• Batman Begins (2005)
• Batman (1989)
• The Dark Knight Rises (2012)
• Iron Man (2008)
• Batman Returns (1992)

[ovidiuch]

See how good "know for" movies for person are

OK, it turns out trakt.tv doesn't really have a rich Person API. You can barely request info about a person, searching them by name http://api.trakt.tv/search/people.json/368b0395732268808fa8d151d2484e8c?query=Wes+Anderson&limit=1, but there's no way to list their associated movies. Nor do most of people have an associated photo.

So far trakt would only be an option if we gave up browsing people altogether.

[ovidiuch]

Let's throw Rotten Tomatoes API in the mix as well: http://developer.rottentomatoes.com/, will update comments

[ovidiuch]

Turns out themoviedb is the only API providing extensive info on People as well, not just Movies. trakt.tv, RT and http://www.omdbapi.com/ as well, don't. Even though trakt.tv and RT seems to have nicer APIs, we'll stick to themoviedb because they make more data available and through the use of a single API call.

[ovidiuch]

Since themoviedb is the winner, only going to mention their rate limit policy:

We do enforce a small amount of rate limiting. Please be aware that should you exceed these limits, you will receive a 503 error.

• 30 requests every 10 seconds per IP
• Maximum 20 simultaneous connections per IP