The 36floors Blog

This week, Facebook, Google and Twitter released their year-in-reviews for 2012. All of them revolved around largely the same topics- The US Presidential elections, the death of Whitney Houston and Gangnam style! This may sound pretty obvious, but it really oversimplifies the bigger picture- 72 hours of video are uploaded to Youtube every minute, half a billion tweets are made on Twitter everyday, and 1.2 trillion Google searches in a year!!

The point we're trying to make is that it isn't just about the amount of information that's already there on the web, it's the rate at which the size of that information is growing. And the growth isn't likely to stop. There's been a 566% increase in the number of internet users since the start of the millennium, and barely a third of the world's population actually has internet access! So, to put it plainly, there's a lot of information being pumped into the internet everyday, and we haven't even hit full throttle yet.

There is, however, one tiny little detail that people forget to mention when they talk about stats like these. Most of this information is time-sensitive. Simply put, most of the content on the web that people are consuming right now, is almost useless a week from now. So then why are there so many different aggregators that are trying to make sense of this information live, and offering you the content available right now, and so few of them (read nobody) archiving all the stuff that actually is worth consuming? 

Don't get us wrong. It's not like we don't keep refreshing our twitter feeds every few minutes, in case we miss out on something, but if you had to build a little internet scrapbook of all the stuff that you'd like to keep and remember, would you really save something like this? While something like that may be interesting to a lot of people, it's not something that people would like to store in their memories. But the fact remains that most news is essentially an unfiltered narrative of what's happening around the world. It lacks value without context. And that's precisely what we're trying to avoid. We want to give you a stream that's filled with interesting stuff, one that can keep refilling itself. Not one where you're either waiting for something new or are bored enough to re-read what was on top yesterday.

And that's our motivation really. 36floors is about sharing the best links from across the web for others to discover and rate them, so that quality is an inherent part of the system.

In the last post, we spoke about how algorithms aren't very good at discovering new stuff and have a tendency to stick to a pattern. In this post, we'll talk about possible solutions to this problem. 

It's actually quite the most obvious thing in the world. You've probably even seen this method in implementation in lots of places. The idea is to use real people instead of algos pretending to be people. There. You're probably shaking your head wondering what's so smart about this method to begin with. Well, here you go. 

Apart from the big disadvantage of filter bubbling that we mentioned last time, algorithms have one other problem. Algorithms are basically incapable of gauging the quality of an article. This may seem odd, but people still haven't found a way to write a crawler that can actually tell you how good an article is. And it's gonna take a really long time for something like that to actually be possible. Quality, at least for writing, is something that is intangible, unlike say quality for factory-manufactured goods. The only way to measure this is by asking people who've read what's being rated. And by doing this on a large scale, there's a good chance that what you've measured is pretty accurate since it probably takes care of any biases that individuals might have.

Using the community to rate content is a method that's almost as old as the world wide web itself. Users submit links and other users vote them up or down into oblivion. Digg and Reddit really pioneered this approach during the 2000s , but they sorta lost their grip with the entry of more versatile social networks like Facebook and Twitter. Users soon switched from "digging" and "burying" links to liking and retweeting links. So clearly, the concept of communities rating links has existed for quite some time. So what's new about it this time around? 

The concept may be cool, but the implementation not as much. The thing is that the most liked links on Facebook or the links on top of Reddit aren't really the best articles or submissions. They're the most popular ones. Which probably explains why Reddit's front page is infested with pictures of cats, and my timeline is filled with memes. Using the same method (upvoting/liking) to express both quality and popularity leads to confusion and ambiguity and over time what results is a dilution of the general quality of content. This becomes an issue since it's now much harder to find the content that you want to read.

Separating quality from popularity is vital to segregating and organizing information on the web. Currently, only Wikipedia and Hacker News (only in the technology niche though) seem like communities that are focussed entirely on quality. Quora seems to be heading there. Sadly, there seems to be no community focussed on sharing and rating the best stuff (not the popular stuff) from across the web. Oh wait, there's one. It's called 36floors.

In recent times, there's been a huge increase in the number of personalized news apps on mobile platforms. Apps like Flipboard and Zite do a great job of learning from your reading habits and consequently feeding you content that they think you'd like. This approach seems to be working, if you look at the number of downloads that these apps get on iOS and Android put together, but there's one tiny problem.

YOU MEAN SODA BUBBLES?

The inherent problem with using algorithms, apart from the creepy fact that they track every single thing that you do, is something known as filter bubbling. A filter bubble is a situation where an algorithm serves you information based on information about you (your location, your interests, etc.) and what results is that you get "locked-in" in your own echo chamber. Simply put, this means you keep getting a lot more of the stuff that you like and barely anything else. This becomes more serious when, say, you Google Japan, and you get information about Japanese tourism, based on things you've searched earlier. Problem is that the rest of the world is watching the nuclear meltdown in Fukushima, Japan on TV. This sort of ignorance occurs a lot more with search engines, and that's something that DuckDuckGo is trying to solve. 

So how does this apply to personalized reading apps you ask? It's almost exactly the same situation. You enter a bunch of interests that you think would be cool and interesting. Over time, it keeps getting filtered and you get served more of the same stuff. Now this isn't really life-threatening or anything, but it sorta misses the point of the internet. The internet was built for connecting, collaboration and discovery, and personalizing all the content that you use consume defeats the purpose. The inherent paradox is that you don't know what you don't know, and filter bubbling only accentuates the ignorance.

Personalizing content isn't necessarily a bad thing. You may have absolutely no interest in reading about weirdly shaped turtles in the Galapagos. But if the entire world knows about the last remaining weirdly-shaped turtle dying, wouldn't you like to be in the know?

                              So, what is the solution then? Well that's what we're working on as well. Do sign up for our Beta here to get notified when it's up and running.

Hi.

Thanks for stopping by here. This is a blog to accompany our site 36floors.com, which is currently being built. You can register for the launch here, so that we can notify you when the site is up. We hope you'll like it.

Our goal is to build a tool for sharing and discovering awesome and relevant content gathered from across the web, which everybody will use. We believe that as the web continues to scale massively in terms of size and diversity of content, the only way for a person to wade through this swamp is by taking the help of the rest of the world. Algorithms are very good at sorting through large quantities of data, but not as good at judging the quality of content. The best way to do this still is (and hopefully always will be) to ask somebody else who's already seen or read the said content.

We believe that the only way to sift through the massive amount of content out there is for people to come together and use their collective wisdom. 36floors is a tool for harnessing the wisdom of the crowd for the crowd.

We hope to do something big, and we hope you stay with us on the awesome journey ahead and help us realize our dreams.