Hacker News new | ask | show | jobs
by tzs 2645 days ago
There are still some oddities.

"How to Train Your Dragon" (2010), 99% with 206 reviews does not make the list, but many older movies, with lower scores and lower number of reviews do. E.g., "The Wizard of Oz" (1939), 98% with 111 reviews, and "It Happened One Night" (1934), 98% with 56 reviews, and "Casablanca" (1942), 97% with 79 reviews, and "Alien" (1979), 97% with 115 reviews.

It looks like the older a film is, the fewer reviews it needs backing its score to make the list.

It's also possible that the age of the review is taken into account, so that a modern review looking back on a film counts more than an older review, because that would probably be more relevant to someone today deciding what older movies to watch.

For example, take "The Good, the Bad, and the Ugly" (1966). It wasn't very well received by critics in 1966. Now, though, it's at 97% on RT, and is regular included in lists of top films (it used to be on the RT top 100, but looks like it has been pushed off).

1 comments

Above all else, RT needs their ranking to be perceived as legitimate. The general public knows there are a handful of old movies that should certainly be in any legitimate Top-100 list. Casablanca, Citizen Kane, The Wizard of Oz, etc. These are movies recognized by the general public as movies that are considered masterpieces. If those movies were not in the list, then the general public would not consider the list legitimate. I'm talking about people who've never seen any of those movies, but are aware of their reputations and will dismiss any Top-100 list that doesn't include those movies.

So the consequence of that is clear. If you are tasked with creating a Top-100 movie list, you better tune whatever algorithm you use to include those movies. You'll butcher your beautiful algorithm to ensure those handful of classic movies always rank high. You might go so far as to put those movies in a special list and inject them arbitrarily into your Top-100 list, circumventing the normal ranking algorithm, because having those particular movies in the Top-100 is a design requirement. It's more important that those movies be in the list than it is for your ranking algorithm be fair and unbiased.