
Flickchart at its core is about two big things: helping you discover and rank movies to develop a coherent chart of your favorites, and mashing everyone’s individual charts into a globe-straddling list of the greatest movies of all time. All the other stuff — social features, recommendations, and so on — flow from these overarching pair of concepts. If either are broken, it’ll ripple through the site in ways small and large.
Today, we’ve upgraded the way we calculate the “best” movies on Flickchart as decided by all of our users (a.k.a. the “global” rankings) to be more accurate than they have been in the past. Previously, a movie had to have an excessively high number of users that had ranked it before it could ascend the global charts – meaning that blockbusters had an unjust monopoly on the Top 100, and as a result, many highly-ranked independent and foreign movies were shut out.
The new algorithm uses a Bayesian formula to take into account those movies that have been ranked by fewer total users, but whose average rankings are very high: several examples currently in the new top 100 are Let the Right One In, The Seven Samurai, Oldboy, and Amélie. Note that this does not use a movie’s winning percentage (how often a movie wins a matchup) but rather the average position where each movie exists on everyone’s individual Flickchart.
In addition, you’ll notice that newly added movies, such as those recently in theaters, will climb the charts much more quickly if they’re ranked highly by our users (e.g. The Social Network, Inception, Harry Potter and the Deathly Hallows: Part One, etc.).
While we think the new overall algorithm is much improved over our previous version, we’ll continue to make minor tweaks as we go along, and we’ll be sure to keep you posted as we do.
A lot of you out there trying to get to the site for some R&R (ranking and relaxation) yesterday were greeted instead with our maintenance page, which promised that we’d have the site back up shortly.
The problem turned out to be more serious than we initially thought — one of the hard drives on our database server had simply disappeared mid-process. Although the drive came back after a server reboot, the database was partially corrupted and inoperable.
We reverted to our most recent backup and could have gone live again after only an hour or two; but that would have meant the loss of about 3 hours of new users, rankings, comments, and so on. We know how frustrating it is to put all that work into your list and discussions and lose it all, so instead of rushing the site back online, we went to work trying to patch up the damage and port over the missing data.
Being a couple of working schmoes, the repairs could only begin in earnest after the day jobs were done, but we’ve been at it since then.
As of half an hour ago, Flickchart is back online with nearly all of the missing data and seems to be cranking along fine. We’ve definitely got a hardware issue going on, and we have several high-priority tickets filed with our hosting company to get it all taken care of. In the meantime, we’ll be working on our logging and backup procedures to handle situations like this better in the future.
Let us know if anything seems amiss, or if data seems wrong or missing. We’ll check it out.
So…although definitions of the word differ, a 9.5-hour timespan — from 3:20pm to 12:45am EDT — probably doesn’t qualify as “shortly.” We’re very sorry for the long downtime — one more bump in the road on the way to our public launch in September.
To all of our users, thanks as always for your patience and understanding.