How Google Ranks Videos

Nirnimesh writes "Google reveals their ranking system for videos on the official blog. The system lists videos according to their country-wide popularity. From the article: 'We use algorithms to identify videos that are suddenly becoming popular, and then rank them based on how popular they are -- and how suddenly they became popular. We've been using this list internally, and now it's ready to share with you, so check it out. Right now this feature highlights videos from close to 40 countries, including Argentina, Australia, Belgium, Brazil, Canada, Estonia, Finland, Greece, Hong Kong, India, Israel, Japan, South Korea, Mexico, and New Zealand, to name a few.'"

Comments

No pigeons?

[MarkByers]

And I was so sure they were using a team of trained pigeons...

Re:No pigeons?

[markild]

And I was so sure they were using a team of trained pigeons...

Nah.. That's just for regular web-pages.

That's what makes this so special. They actually had to sit down and make an algorithm.

Barbie Girl

[rogerramrod]

I don't think the algorithm is quite ready for prime time yet,
seeing that currently the most popular video on Google is of a chubby nerdy tranvestite playbacking a song of Aqua

I wonder how accurate this is.

[micheas]

If you look at what google publishes about pagerank and compare what they say with the search results returned by google.com you will notice discrepencies.

For example google claims that they do not return urls that contain '&id=' but that is clearly not the case. (Joomla and Mambo sites without seo enabled would have substantially fewer pages returned than they do if this was the case.)

I have come to feel that I can trust google about like I can trust my own contries military. (after independent verifiaction, and I need some reason to believe second source is not compromised.) But there are gems in the statements so I read them and after I am done reading I have more questions than before.

This looks like a small glimps into one component of pagerank. But the article is pretty light on substance.

Nothing like destroying an early beta

[baadger]

So they rank video's by how _suddenly_ they become popular...and we go an put it up for a Slashdotting. Good luck unraveling them stats Googlers.

I'm flabergasted

[suv4x4]

You know, maybe it's not Google who hypes this up. Maybe they just posted it in a blog and the community took it up.

Every single little thing Google does, no matter how trivial, is reported to take down Microsoft and take over the world.

But COME ON, is this really worth an article on Slashdot:

[we] rank them based on how popular they are

Shit they better patent it before someone else figures it out!
It's only literally every site with plenty of items that can be sorted based on popularity.

Re:I'm flabergasted

[suv4x4]

We are geeks... I mean, check out what is "news" to the rest of the world. A dumb blond wrecking her car...

I find this especially ironic. People care about Paris Hilton's car crash not because car crashes are particularly rare, but because Paris Hilton is really popular.

And you're reasoning that Google ordering items by popularity is very important, not because ordering by popularity is particularly rare, but because Google is really popular.

But you can always find comfort in the idea that we're "the geeks" and "the rest of the world" is just plain dumb and undeserving attention.

I've never been so embarrassed in my life.

[born_to_live_forever]

I'm from Denmark, so naturally my first thought was to see which vids were considered most popular in Denmark, according to Google. The results were disheartening, embarrassing, but far from surprising.

Crazy Frog.

Dozens and dozens of crappy variations of a done-to-death meme.

Kill me now, and get it over with.

Re:How hard could it possibly be to...

[moonbender]

That's not hard. That's also not what they do, apparently. They actually try to detect popular videos before they are popular. That way a current video gets a higher rank than an old fad that got 2 billion downloads over the years. Think measuring acceleration instead of speed. That's probably not very hard, either. But the resulting page is still pretty cool.