David Pakman's Blog www.pakman.com

Facebook is No Longer Real-Time

facebook expand
Mar 07

It’s a good thing Facebook is thinking of redesigning the News Feed. Because I think a funny thing is happening to Facebook. For me, the news feed no longer surfaces anything of interest. The opaque algorithm behind it is just not able to produce anything relevant and, more important, timely, at least to me. Facebook appears to be turning back into what it once was: a way to research people in non-real time. A look back into the past. A people-stalking product. It’s back to being a personal LinkedIn.

People publish stuff on the (increasingly mobile) web that is timely and relevant. Sharing baby pictures isn’t really one of those. Sharing pics of how you are experiencing life, which is the Instagram use case, is a great example of this. But my News Feed does not have anything like that in it. My Instagram feed does.

People share highly informative and timely links to news articles and blog posts on Twitter all day long. But my News Feed does not contain any of those. And when I share these types of posts on Facebook, I get no engagement. When I share pics of my kids, I get a lot.

People share bookmarks of products and apparel they want to buy on Pinterest all day long. People don’t do that on Facebook.

Facebook started as a non-real-time service. It was a way to check people out. In the face of the rise of Twitter, they responded aggressively with a News Feed product that showed promise. But now I feel they really screwed the filters up that govern that feed, which creates feedback to those of us who post into it and it feels like a vast river of noise and irrelevant posts from people and events who aren’t really relevant to me. Perhaps most importantly, I can’t tune it. The tuning mechanisms are either too subtle (“hide”) or too crude (“report as spam”). I feel powerless.

The irony is that LinkedIn is moving to increase daily engagement by syndicating highly informative posts from influencers. They are trying to become more real-time just as Facebook seems less so.

It’s still amazing for stalking people, though.

Author David Pakman
Category Social Media
Comments No Comments

Not All Traffic Is Created Equal

Sep 26

To build the online media giants of tomorrow, companies need models where the costs of both content and distribution are near zero. Google, Facebook, Twitter, Instagram, Pinterest and countless others employ this model. These models allow scale to emerge at very low-cost.
And in these particular examples, the scale achieved is astronomical — on the order of hundreds of millions or billions of users. In thinking through how to build businesses around this scale, a lens emerges: what kind of traffic produces that scale?
In the case of social media companies like Facebook, Twitter, Instagram, Pinterest and Tumblr, the root activity on the site is the sharing of content. But the content shared on those sites differs widely, particularly around which content attracts the most engagement. Broadly, Facebook attracts photo sharing and light-hearted personal content. Twitter responds far better to true news and topical information sharing. Tumblr seems to resonate around entertainment and creative media. And Pinterest lights up around home design, apparel, food and other commercial items. (I am taking some liberties by generalizing, but you get the point.)

At the scale of Facebook, you could have your users share almost anything and still be able to build a large business, purely by loading the site up with lots of advertising that is (at very least) rudimentary targeted. At that scale, you can reach billions of dollars in revenue. And I believe, even at their scale, their ad load will need to further increase (along with their targeting abilities) in order to signficantly grow the business. (They also must move advertising off-site, as they are now doing, which I detail in this post.)

But if your service attracts particular verticals of content engagement, not all content is created equal, and some is much more valuable than others.

I divide traffic/content engagement into three buckets: topical, informational and transactional.

  • Topical content engagement is what is mostly taking place on Facebook, Tumblr and Twitter. It is comprised of posts generally linking to news, information, family, entertainment, photos, etc. The signal in this stream is the lowest of the three in terms of monetizeable traffic.
  • Informational content, often found on sites like SlideShare, Zillow and automotive blogs is the sharing of information that is near the top of the funnel for demand creation. Things like business white-papers or product reviews are perfect examples of informational traffic. This traffic has significantly more value than topical traffic, and excels at attracting endemic advertisers in the key verticals of travel, auto, tech, financial services, real estate and pharma, to name a few. Intent is well understood in this traffic and the signal is strong.
  • Transactional content is traffic that is essentially one click away from a purchase. Obviously, traffic found on ecommerce sites is the prime example of this and search traffic is a close second, but increasingly Pinterest is proving itself to be a massive source of high-converting traffic. Here, intent is clear and the signal is strongest.

I believe, with the Facebook share price correction, we are entering a period where sites based on topical content traffic are going to struggle in generating value for themselves. Much of the valuations around the consumer web are rationalizing, and because of that, investors are once again focused on understanding business models. Social media properties building traffic around informational or transactional content will be significantly more valuable than topical ones in this forthcoming period. This general notion that every social property with scale will be able to create their own custom “social ad” units and monetize themselves consistent with their earlier valuations, I think, is flawed, unless those properties are in the two higher tiers of content.

Got Klout?

klout-logo-full-color expand
Jan 04

Marissa and I are pleased to discuss our latest investment (and our first one as a team). We are excited to join with fellow firms Kleiner Perkins and IVP in investing in the internet’s standard for measuring influence, Klout. As the internet moves from pages to people, Joe Fernandez‘s vision of the need for “Pagerank for People” is spot on. Klout’s algorithms score the actual influence of people as they share on the social web. They attempt to measure your influence by observing interactions on the social web. As we all work to build and manage our online identity and profile, Klout helps measure our reach and topics of influence.

In every other mass media, measurement provides a benefit to the advertisers who subsidize that media. Large companies have emerged based on, frankly, less than perfect measurements systems. In TV and radio, panel-based inference measurement somehow have passed as a legitimate way for advertisers to make decisions on where to spend billions in advertising. These incumbent measurement firms became standards for measurement within their domains. Klout has the benefit of being able to measure actual data, not inferred data. They aim to score the entire social web. They currently have scored more than 300 million users and are scoring and re-scoring a mind-boggling amount every day. With more than one billion people on the social web today, they are by no means complete. Nor are their algorithms perfected. Just as Google changes their PageRank algorithms hundreds of times a year, Klout will evolve their data science as the social web changes to provide the most accurate influence scoring on the web.

Klout has the distinction of being one of the few companies whose monetization plans actually benefit its users. Using Klout to identify influencers in particular topics, brands offer new products or special “Klout Perks” to you in the hope that you will like them and share your point of view with friends and followers. This relationship, unlike interrupt-driven advertising, benefits both parties. Klout has worked with more than 100 brands like Starbucks, Audi, Spotify and Microsoft and has hundreds more lined up to do the same. Joe speaks infectuously about his plans for taking Klout to “the real world”. He imagines restaurants knowing your Klout score when you call to reserve a table, airlines printing your Klout score on your boarding pass, and of course call centers knowing your Klout score when you call to complain. Already hotels are using your Klout score when you check in to decide upgrade policies.

Aside from this exciting vision and stellar progress, two other themes draw us to Klout. One, we hold a passion around seeing the relationship between a brand and a customers changed. We believe that the social web requires brands to respect us more. To take our point of view more seriously. To adopt policies consistent with good service and fair treatment. No human should have to sit on a plane for seven hours on the tarmac, of course. But also, utility companies should be held accountable for poor service, cable companies should be held accountable when we stay home from work for a day and the repair crew never shows up. Banks should be called out for imposing hidden fees in the dark of the night. And finally, our governments and elected officials should hear from more of us more often. In this age of declining influence of traditional media, Klout enables our individual voices to be more influencial with instutions who hold power. That is exciting to us.

And finally, Klout supports our view that we are shifting from an attention economy to a data economy. The last ten years of digital media on the web have been built on attention. Those web properties that amassed our attention (generally by stealing our eyeballs away from traditional media) and reached scale have been rewarded with great businesses. Yahoo! got our attention with email. Google got our attention with great search. Facebook gets our attention with photo sharing. We believe the next ten years will be built around data, and in particular, social data. We have invested in M6D for its leadership in social ad targeting. We invested in Singly for its leadership in building a social data locker and app platform. And now we are investors in Klout for its leadership in social influence measurement. We salute Joe and his team for amazing progress so far, and are pleased to be along for the ride.

Is Social Data The Next Investing Frontier?

Oct 19

Much of the excitement around internet startups over the last five years has been around social services. From Facebook to Foursquare, from Twitter to Instagram, from Yammer to Zynga, significant investment dollars and entrepreneurial effort has gone towards capitalizing on the fact that we are all linked together by connected devices. These connections present great opportunity to disrupt the traditional ways of attacking markets like shopping, travel, communications, media consumption, gaming, etc. There are plenty of other big investment themes, of course, like local commerce (Groupon) and cloud services (Cloudflare and Dropbox), but social has been the dominant theme. The first wave of social companies were social utilities and social media (including gaming).

I believe that is shifting and has been for some time. Other agree. We have been pursing alternate investment themes these past few years and the largest recurring theme for us has been data. This is also not a new theme, but it is growing in prominence and awareness, punctuated by this week’s Web 2.0 Summit whose theme is “The Data Frame”. We have invested deeply in data-based businesses whose efficiencies disrupt their less-efficient or less powerful legacy brethern. AdTech is one such area. Healthcare is another. Payments is a third. Security is a fourth. And soon, the consumer web is likely to be further transformed by businesses based not on social utility, but on social data.

Plenty of consumer startups use data to make product decisions. That is not what this post is about. It is about consumer businesses actually based on the value of our individual social data. Through the use of so many exciting social utilities, we are creating more data about ourselves at an increasing rate. This data becomes more valuable to us when developers can access it in an aggregated and trustworthy way.

Today, an investment we seeded back in March called Singly is making its intentions known at Web 2. Their vision is audacious; individuals must be in control of their social data. I blogged a little bit about this opportunity here. Today Singly emerges as a developer platform to bring that vision to reality. John Battelle blogs about it here. I think their emergence shines a light on the investment opportunities around social data as well as the opportunities to launch open personal data platforms.

Jeremie Miller, Singly’s Co-Founder will present today, Wednesday October 19 at 2:20pm PT/5:30pm ET. You can catch the livestream here.

The Magic of Minecraft

Sep 01

On the Saturday night before Father’s Day, I called my three kids together and asked them if they’d like to earn an extra five dollars. My wife and I were having a dinner party at our home that evening, and if they agreed to take baths/showers themselves, put on their jammies, brush teeth and settle into bed themselves, they could earn the extra money. Highly motivated, they took the challenge and went to bed without parental involvement. The next morning, after awarding them their well-earned compensation, the older two (ten and eight years old) immediately asked if they could buy the paid version of Minecraft with their money. And from that moment, my eyes have been opened to a magical creation.

A view of a Minecraft world

What is Minecraft? It is a java-based web computer game the kids had been playing for a few weeks. The free version allows them to play a primitive version of the game in single-player mode. I checked out the paid version and upgraded them each for €15. The game was created by a mysterious and revered Swedish indie game developer named Markus Persson, known as Notch (@notch) and his team of 8 others called Mojang. As my kids showed me the somewhat crudely drawn lego-inspired world of trees, grass, oceans, islands, zombies, spiders, skeletons and the dreaded creepers, I was intrigued. They were mining for ore, collecting supplies, crafting new items with pre-determined recipes and sharing their learnings. There were no spaceships, no lasers, no bullets, no armies, and no blood. In place of the fast-twitch first-person-shooter games dominating console and PC gaming was a construction oriented world set in primitive times that has captured the imagination of about 10 million free users and 3 million paid users worldwide. (Yep, that’s more than $66M in revenue in less than two years.)

The dreaded Creepers!

Watching them play in parallel on two different computers, I assumed there was a multi-player version. After some googling, I found literally thousands of multiplayer serversrun entirely independently from Mojang. We tried one and found very rich worlds with scores of simultaneous players and lots of rules. Not feeling advanced enough to join these evolved worlds, some googling brought me to a free java version of the server. It was Father’s Day after all and I’d rather be playing with my kids than not, so I launched a local server in our house. It worked like a charm. We all logged in and then the magic really started. We were now playing in the same world, chatting with each other, banding together to mine, build and defend our creations. After a few hours glued to our computers and to each other, it was clear we were going to be playing this for a long time. I was flying to California that night and thought this would be a great way to keep in touch with the kids, so from the car on the way to the airport I spun up a Rackspace linux box (Ubuntu, of course), installed java, and brought the server up. I made some DNS changes to the pakman.com domain name and launched our server more publicly. It would now be possible for us to play together no matter where we all were. Quickly addicted to the tasks of mining and building, I awoke at 4am California time each morning to play with my kids online for an hour before they left for school and I left for meetings. At night I’d check out what they made. They wanted to play Minecraft every waking hour of the day. And so did I.

Creative City on the Pakman Minecraft Server

Fast forward to today. The three of us have played probably more than 200 hours of this game, mostly together. We pray for bad weather on a Saturday to cancel tennis or other outdoor commitments so we can build and explore more of our Minecraft world. My kids have invited many of their friends, almost all of whom were already playing Minecraft, to join our server. We have more than 30 kids who have tried our world and at least 4 kids on at any waking moment of the day. I have consulted other Minecraft server Operators (“Op”, for short) and become a sophisticated Op myself. I upgraded the server to an 8GB quad core box to allow more simultaneous players. I moved the world onto a RAM disk to accelerate the delivery of graphic chunks to the clients. I now wrap our server in the community-created Craft-Bukkit framework to allow me to add and modify server mods without bringing the server down (the kids hate downtime). I added an economy to our world so the kids can buy, sell and trade items in exchange for money. I added a bunch of NPCs (non-player characters) to richen the world experience. I added a Group structure. New players come in as guests with limited abilities so they cannot trash the world (“griefing” in Minecraft parlance) until we trust them and know them. We even added an alternate world called The Creative where kids are encouraged to build elaborate structures. These kids created an entire town complete with a church, fire station, castles, restaurant, airport, farm, houses and a library (okay, I made the library).

The mercurial and revered Notch

A few weeks into the experience, I got a frantic call at work. Some kids had come into the server and were destroying homes and killing players. “Dad, quick, you have to do something. You have to ban these kids from our server!” So, I banned a few of the wrong-doers. It may not surprise you to find out that the few who were banned were already somewhat known as the trouble-makers at school. Now we have griefing-protection tools and anti-cheat technology on the server to help bring a little order to the world. Not too much, but just enough to keep the community healthy. What is happening here? First, it is important to understand that Minecraft is not just a game. Although known as a “sandbox” 3D construction game where users create in a virtual world with basic rules and logic that determine the way the world operates, Minecraft is a true phenomenon. Head over to YouTube to see this first hand. There are more than one million videos uploaded by gamers showing off their creations, tips and ways to mod the game. In this video that made its way around twitter a few weeks ago, a group in the UK created one of the most elaborate looking dams I have ever seen. In another one, a group on a server created a Happy Birthday message to Notch. Most extreme? This block for block replica of the Starship Enterprise or the Arc de Triomphe. The game is actually still in beta, the server is buggy and there is very limited developer support from Mojang. Despite this, there are tens of thousands of developers who have written mods and plugins, hundreds of thousands of skins and texture packs to alter the look of the game, and many community wikis and forums with hundreds of thousands of posts and articles. (It’s not particularly easy to mod the game without a nice API…these devs are disassembling java code and hacking it to make the game work differently.) Unconvinced? Watch this Best of Minecraft 2010creations video.

Arc de Triomphe, Pyramid and the Parthenon

Notch and his unbelievably gifted team at Mojang have unlocked an enormous reservoir of creativity largely among kids. I was not too surprised to find my ten year old’s teacher allows the kids to play Minecraft in the classroom to teach construction and encourage creativity. But more than that, I am observing first hand how the players develop ad hoc rules for social interaction in these worlds. This is so much more than a game. This is the inevitable progression from one-dimensional social networks like Facebook to virtual world social networks. If the Mojang folks supported a more robust server architecture and possibly larger game maps, we could see worlds with hundreds of thousands of simultaneous players. I believe Minecraft fulfills the promise Second Life and IMVU have not; these players are not waking up and deciding to go into a virtual world. They are deciding to play and build in Minecraft and the world and social rules follow from that. Minecraft gives its players a reason to come together to interact, much like an outdoor BBQ brings us together to eat and socialize or a dance club brings us together to dance and socialize. Minecraft also presents a number of challenges to traditional video gaming in general. One of the reasons I believe kids love it is because every single block in the game is moveable and alterable. The entire game landscape can be redrawn by the players, one block at a time. This is enormously empowering to a child who lives within a strict set rules about what may and may not be touched in the real world. In Minecraft, you can touch everything. (The blocks do adhere to primitive logical rules like gravity and the effects of states of matter, so it is not a complete free for all.) In addition, the marvel of the game’s success cannot be understated. It has not even been formally released and it has 10M players? And it was developed by a tiny team (relative to big game development) who built and then leveraged a rabid community of their users, many of whom are technical enough to hack and improve the game in all sorts of unimaginable ways. So where can this all go? If the team at Mojang wanted to and thought this way, I think this game could be a platform for global social interactions and easily become the largest virtual world social network. I can see this reaching 100M players. They could more formally support the developer and multi-player server interfaces to really let the game be extended in more reliable ways. They could allow for different world generation algorithms to be used to create more variety in the basics of the map structure (which could unlock a different set of creativity). My friend (XMPP, Telehash and Singly co-founder) Jeremie Miller excitedly hopes for an ability to teleport among various servers without re-starting the game. This would require intra-world permanence of your items and state but would allow people to move from community to community very quickly. As constructed today, each client actually can create and run its own single-player game. Why not allow every client to be a server and host additional players? If they used Telehash, Jeremie points out that “anyone can portal to any other running world on any computer anywhere in the world.  Any server you’re on you can always build a “home” teleport to a world on your computer, as well as build portals from yours to all your favorite multi-player servers.” This is an exciting vision. Jeremie also suggests allowing media assets to be delivered from the server to client, currently not permitted. The only way for new characters and scenery to be introduced is to simultaneously mod both client and server. Allowing the server to add new elements to the client would obviate the need for all users to upgrade their clients just to receive new game items. This all being said, I wouldn’t change much. Ecosystems like this are fragile and are very hard to get right. Notch and his crew have gotten it pretty much perfect as it grows organically every day. I truly believe this team is quite genius. The amount of thought that went into getting this balance just right to encourage us to explore and learn on our own and then want to share our learnings is staggering. This week I purchased three tickets for me and my two kids to attend MineCon in Vegas in Novemeber, a community-created convention when the game will be officially released. Wanna come? If you’d like to try out the Pakman Minecraft Server, please send me an email and I will happily send you the address.