Today is a big day for us. We’re proud to announce the launch of piki on iPhone!
Piki is based on a really simple idea. The best recommendations for new music come from our friends. We all have those 2 or 3 friends that always seem to have the best new music. With piki, we allow you to follow…
When some people say you should learn to code, they usually summon up imagery of start-ups and iPhone apps. But that’s just part of why you should learn to code. Learning to code gives you another tool to examine life.
I recently had a random thought. I wondered how long are the most viewed YouTube videos? Without coding, this would take awhile to figure out. It would involve a lot of browsing and data entry. It probably isn’t worth the time to figure out.
With coding, it took me less than 15 minutes (and another hour to polish it enough for this blog post and write this). It’s not a lot of code either. You can view it here.
I learned some interesting insights from playing around with YouTube’s API.
First, here’s a scatter-plot of the top 950 videos on YouTube. The y-axis is the number of views a video has had and the x-axis is the duration of the video (in seconds).
First off, the equivalent of 10% of the world has watched Justin Bieber - Baby. There’s also a huge gap between that video and the still hefty videos at 450 million views.
Aside from the interesting outlying datapoints, the main thing that immediately jumps out is a wall of points right around 4 minutes. I started wondering why 4 minutes? If Bieber isn’t enough of a clue, it has to do with all the music videos on YouTube.
I decided to remove all the datapoints that had a category of music and got this.
Again, there’s this huge gap between the leader (Charlie bit my finger) and the rest of the videos. My guess is that the top video is a self fulfilling prophecy and people keep watching it because it’s the top video.
Now we can see that wall at 4 minutes disappear, but there is another less defined wall at around 2 minutes. This feels about right based on my own viewing habits. If something is longer than 2 minutes, I’m less likely to watch it (blame the internet for my attention deficiency).
For those curious, here is the music only graph
If you had a strange sense that the length of pop songs are generally the same, this graph shows it.
Therefore, if you are going to make a music video, try to keep it to 4 minutes and if you are going to make a comedy video, try to keep it under 2.5 minutes.
Learning to Code
If you’re interesting in learning to code, there’s a great site called Codecademy that walks you through from complete beginner to beyond. Anyone can learn to code, the trick is don’t get overwhelmed and start with the basics.
For those of you interested in the code, here they are (warning: you will get sucked into hours of watching videos):
The code (written in python and requires cjson, but can easily be changed to use simplejson)
One of Facebook’s core philosophies is that a user’s privacy is paramount. Part of that privacy is the data you create by browsing Facebook is never shown to other users. You’ll never see a newsfeed story that says, “Joe viewed your profile 1,024 times.” It’s that most basic line in the sand that has helped Facebook grow like it did. It has allowed us to all breathe easily when we view people’s profiles, photos, timeline, etc. We’ve all done it. Maybe it’s a potential girlfriend or an ex-boyfriend or a stockpile of images from a party. One of the most important parts of Facebook is that we can all browse it freely and without consequence. When we decide to take an action such as commenting on a photo, it is then that we make ourselves known. The simplest action being, I like this.
The current Facebook newsfeed is a whole different story. It’s littered with things people have listened to, watched, read, and bought on other websites. Many of my friends don’t realize these are showing up. When I tell them they may not want, “Young girls dancing on the beach (via socialcam),” showing up, they are thankful and ask me how to delete it and stop it from happening again. I had another friend read an article, “5 Weird Things That Make You Horny (via mens health).” Even though she might be perfectly curious, it reads a bit differently when it’s on your newsfeed. A user is free to turn these off, but the default is on and anyone who develops software knows that default is king.
These two philosophies are completely contrasting views of the world. Facebook has for a long time been about explicit actions. When they introduced a like button for websites, this was a great example of an explicit action. I click a button and tell the world I like this page.
The implicit version of this would be every website you view gets sent to Facebook and shows up in your feed. It’s not hard to imagine how that would be received worldwide. What is going on right now is the first step to that reality. If Facebook is moving towards being a browser for external content it should treat itself like a browser. You will never see Chrome post all your viewed websites to Google Plus.
Sites can grow virally without it.
Instagram is an example of doing it right. They make it incredibly easy to post to twitter and Facebook. I use this feature often. Sometimes I choose to take a photo with Instagram simply because it’s the fastest way to get it onto my Facebook and twitter feeds. But Instagram has never implicitly pushed these actions to external social networks and it certainly had no trouble growing.
When we launched turntable.fm, I also drew the line in the sand that we would only have explicit actions. I received a lot of pressure from some advisors that we should be as spammy as possible because it works. We grew fantastically well based simply on having a great idea and product and didn’t have to resort to it. Then, like many other startups, I bent that rule when Facebook introduced the custom open graph. It was a judgment call and came down to a belief that we had to do it to properly compete with everyone else doing it. However, starting this week, new turntable users need to turn on implicit Facebook sharing in their settings.
Can it help sites grow? Yes, but they are viewed as spammy and don’t need it if you build something great.
What’s so bad with implicit actions?
How our friends and peers view us is important to us all. We all spend a lot of time filtering what we say, what we decide to do, who we decide to be with and where we decide to go. We spend even more time deciding which of those actions we want to call public. Implicit actions are a jarring change to this normal flow of how and when we decide to make ourselves public.
Implicit actions require you to behave differently. Every time you do something that might potentially show up on Facebook, you have to think to yourself, do I want this to appear on Facebook? You also need to think about who will see it. Our network of friends is expanding into all parts of life and I am now friends with young cousins, coworkers, the person who cuts my hair, and various other people. You may even have forgotten that you gave permissions to a specific website, visit a few months later and get something auto-posted to your timeline. It’s also a complete mystery of when and what actions will get posted (the onus is on other sites to tell you, which they mostly never do).
Facebook isn’t the only service with implicit actions. I love Path, but did not appreciate when it automatically posted to everyone that I woke up when I opened the app. Path also shows when people viewed my profile (and when I view theirs). Both of these implicit actions make me browse Path less than I normally would.
One of Zuck’s slides during the last F8 was a huge Facebook popup box requesting a user share their most current achievement (getting a mushroom) in Super Mario Brothers. He wanted to fix the problem of constantly being asked by every app if they want to post to Facebook. It’s definitely an issue, but implicit actions are not the correct fix.
There are a few steps that I think Facebook could take to fix this:
- Make a social contract with all users that all browsing data will never be shared and posting will always be explicit (and bask in the collective sigh of relief across the world).
- Make an API for one-click actions that do not require bulky popups. This would be much like a Facebook like button, but something with a customizable action.
- If Facebook must have implicit actions, make them all go into a holding pen somewhere on the site where users can then publish the ones they want (thereby turning them explicit).
The fix is not more user control over when we implicitly share (or by setting permissions to viewable ‘only by me’). That only fixes the problem for myself. It doesn’t fix the problem for my friends, family, and coworkers, many of which don’t realize there is a spyglass into their browsing habits.