Recently, News Corp. has been making a lot of noise about removing their web properties from Google’s search index. This appears to be a desperate move being made by a company that has not adequately embraced the culture or technology of the Internet. The important aspect of a total removal from Google that is not being discussed is how News Corp. would potentially screw thousands of businesses that have outbound links from their web properties.

Partners

Google uses a concept of “trust” when they assemble their search index. In theory, if a popular and trusted site like The Wall Street Journal (a News Corp. company) links to a site, Google is going to pass link juice or link equity to that site. The theory is that if the WSJ trusts you, then you are probably an OK site (i.e., not full of spam and the content quality is similar). The concept of link equity is similar to the professional references provided to a company when applying for a job.

Future

It appears that News Corp. is an old company and their business/negotiation tactics seem to be based on an outdated machiavellian business philosophy. The future is calling you News Corp., evolve or perish.

Recently, we looked at the Skype sale drama, and now it appears as if there may be a shift in the deal. Both the NYTimes and GigaOM are reporting that a settlement may soon be reached between Skype co-founders, Janus Friis and Niklas Zennstrom and the private investors acquiring the 65% stake from eBay. The odd men out appear to be Mike Volpi and London-based Index Ventures. Given the legal circumstances, swapping Index Ventures for Atomico Ventures was probably an easy decision for Silverlake Partners, CPP and Andreessen Horowitz.

Recently, it was announced that eBay sold 65% of Skype to a consortium of private investors, led by Silver Lake for $2.75 billion. At first, this deal sounded great considering eBay purchased Skype for $2.6 billion in 2005 and was never was able to realize the original goal of integrating the product into its online auction service. However, more information has come out that makes this look like a potentially enormous mistake for the new investors. When eBay (at the time led by Meg Whitman) purchased Skype from Niklas Zennström and Janus Friis, Whitman failed to secure the rights to core P2P intellectual property. Failing to secure the core IP, known as the Global Index Software (GI Software) was an enormous mistake that potentially borders on negligence. Whitman,  who is now interested in becoming the Governor of California, may feel the sting of this mistake as people question her ability to properly manage the negotiation of complex deals.

Global Index Software

The core P2P IP (the GI Software) in Skype is still owned by Zennström and Friis in a holding company known as Joltid. Their company has licensed the GI Software to several companies including one of their latest ventures Joost, a video distribution company. Until recently, the CEO of Joost was former Cisco executive Mike Volpi. While at Joost, Volpi led the company away from a P2P approach (using the GI Software) and towards a CDN or semi-centralized content distribution architecture. This was a relatively smart move on Volpi’s part because of the limited upload speeds available to most home Internet connections. Limited upload speeds mean that it takes a lot of dormant/idle computers to serve HD content to a single user. Without an enormous install base, the Joost P2P product appeared to have difficulty streaming high definition content. While leading the migration away from the P2P framework, the Joost engineers had access to the GI Software source code. The transition was not trivial and it cost Joost a lot of time/money. More importantly, this gave Volpi experience in leading a migration away from the same proprietary software in Skype, which — thanks to eBay — is not owned by the company. Not too long ago, we wrote a post about Skype’s Achilles’ heel in their P2P core. Undoubtedly, we are not the only people who see the long-term issues with the P2P core.

Focus on Volpi

The major issues for Silver Lake are the new lawsuits against them and the people involved in the deal. Recently, Volpi’s employment at Joost was terminated. In a new lawsuit filed by Zennström and Friis, they alledge that Volpi (while still an employee of Joost) used his proprietary knowledge of the GI Software (and how to migrate away from it) to shop a deal around to investors to purchase Skype. They also claim that he recruited Joost engineers who worked on Joost’s migration away from the GI Software to Skype. Of course there are always two sides (at least) to every story, but the lawsuit against Volpi looks incredibly damning. This of course is up to the courts to decide but the impact on Volpi’s reputation may extend a lot further than this deal.

Future

We hope all this will not impact the availability of Skype. We use the product a lot but are concerned that the company/product may not weather these times well, due to its current legal woes. Perhaps the smartest thing would be to pay Zennström and Friis to go away, but we fear their price may be too high.

Happy Talk Like A Pirate Day!

In a previous post, we discussed the history of Oracle and MySQL. In this post we will focus on Oracle’s relationship with its new star, Java.

History

Oracle made its first big move into the Java space in 2001 when it acquired a non-exclusive license for the source code of Sweden-based IronFlare AB’s Orion Application Server to serve as the foundation for the Oracle Application Server. Oracle continued to make progress in the J2EE space for the next few years, but it was always stuck behind JBoss, WebLogic, WepShere and others.  Rumors circulated around BEA Systems for years that Oracle was interested in purchasing the company until 2008 when Oracle finally acquired the company. With the BEA acquisition, Oracle came into control of a superior Java Application Server (WebLogic) and one of the few really good Java virtual machines, JRockit. The acquisition of Sun and, subsequently, Java puts Oracle firmly in control of the Java industry.

Future

The majority of the Java virtual machine was open sourced under the GPL V2 license in 2007 via the OpenJDK project. Several libraries related to the Java 2D APIs were withheld because of previous license agreements Sun had made. If Oracle decides to stop supporting free and open source Java, there will be a void in the industry that would require Google, IBM, HP and others to invest heavily in the language to keep dominance away from Oracle.

We expect IBM to make a play for Red Hat before they lose too much of its competitive edge in the software infrastructure industry.

Relevance

According to the May 2009 TIOBE Programming Community Index, Java comprises the dominant 19.537% of the programming language mindset.  The ownership of the Java space gives Oracle a problematic amount of control and influence over the Java platform. As server operating systems become less relevant in favor of logic embedded in platform independent Java, Oracle is well positioned to place extreme pressure on all of Microsoft’s server solutions. Additionally, most of the modern scripting languages have been embedded into the Java virtual machine so the platform truly transcends Linux, Windows, Mac OSX, Solaris etc. Google’s new operating system, Android, is based on Linux but the APIs exposed to developers are primarily Java-based. Additionally, Google also recently released Java support for its platform solution, App Engine, and its Web Toolkit is also based on Java.

With an uneven power shift over the Java platform, it is completely possible that innovation of Java will not continue at its current pace.

Growth

We would like to see Google participate more in funding, development and maintenance of the OpenJDK project to help ensure innovation is pushed forward and Oracle is kept at a safe distance from discontinuing open source releases.

Recently, after a failed bid by IBM, Sun announced that they will be selling the company to Oracle for $7.4B. At Deft Labs, we saw the writing on the wall for this deal in January of 2008. This post examines the historical relationship between Oracle and MySQL and looks at the potential impact on the immensely popular open source database.

History

Innobase, the Finnish software company responsible for InnoDB (one of the storage engines available in MySQL) was sold to Oracle in 2005. Oracle made another move against MySQL in 2006 when they purchased Sleepycat, the company responsible for Berkeley DB,  the low-level open source software which Innobase built InnoDB on top of. Oracle also attempted to purchase MySQL in 2006 but then CEO Marten Mickos told reporters, “We will be part of a larger company, but it will be called MySQL.” It is no wonder Mickos turned down the offer. Most open source advocates probably would have seen an Oracle deal as a betrayal. MySQL continued to progress until they sold to the open source-friendly Sun in January of 2008.

Future

We do not forsee any changes in the status quo for the next year or two. After that, we expect to see more of Oracle’s proprietary database technology in MySQL (which will not be available in the open source/developer edition).

Relevance

Several so-called “web scale” databases are beginning to mature. Most of these databases are document-oriented and typically reduce the amount of logic that can be executed in the database/storage layer in favor of more modern techniques which process data in the application layer. A few of the “web scale” databases available are:

• MongoDB (Open source)
• Amazon SimpleDB (Proprietary)
• HBase (Open source)
• Hyptertable (Open source)
• CouchDB (Open source)
• App Engine’s BigTable (Proprietary)

In a nutshell, with the migration to a new database paradigm, the loss of MySQL to Oracle will probably not be the worst thing in the world.

Recently, it was announced that Microsoft is suing GPS device manufacturer, TomTom over GPS patent violations. TomTom runs on the open source Linux operating system and a lot of people believe this is an attempt by Microsoft to assert their perceived intellectual property rights over Linux.

Normally, this news would not be too exciting, but this time it appears as if Microsoft may be up to their old tricks again. It is widely rumored that several netbook manufacturers have plans to ship devices with Google’s Linux-based operating system, Android. If Microsoft is able to scare netbook manufacturers away from shipping devices with Linux then they may be able to maintain Windows’ stranglehold as people migrate to inexpensive mobile laptops.

Eric Schmidt, Google’s CEO, said, “netbook” three times today during his presentation at the Morgan Stanley Technology Conference.

Note: We have to give credit to Google’s sense of humor. A week before the Microsoft vs. TomTom news broke, Google announced that they would go to any length to fight off patent trolls.

If you examine the pressure on the content distribution industry by top tier ISPs, it’s clear this industry is changing. ISPs are now competing against their transit and hosting customers for content distribution business. This is a fairly bold move for ISPs, but it’s clear that they’re starting to question where their next spurt of growth lies.

Internet Service Providers

In 2006, Savvis (in what may have been a fatal mistake) sold their CDN business to Level 3 for $125 million. Apparently, AT&T also got the memo and rapidly deployed a CDN.

Globally, the majority of ISPs don’t have a CDN offering and they will continue to lose business to those that offer a “more complete stack.” All of this inevitable consolidation is necessary as the content distribution business becomes more efficient. Panther, one of the commodity CDN players offers a CDN solution for ISPs, and it’s possible that this model will become more prevalent because the majority of ISPs are ill-equipped to provide sophisticated (and cost-effective) CDN services.

Why?

Now, why does Google need to purchase Akamai? If the obvious continues to play out then the ISPs will have a renewed sense of “hand” when dealing with their customers. If Google is serious about becoming the dominant cloud provider then they will need all the leverage they can get when dealing with their peering partners. It’s widely believed that Google owns some of its fiber infrastructure and they have been busy building data centers around the globe.

Legal

The good part about Google acquiring Akamai is that there is solid competition in the CDN space. Akamai is the dominant player, but based on their recent stock performance, it’s clear that the ISPs and commodity players are gaining ground.

We don’t believe a legal protest of this acquisition would have much credibility. Akamai also employs a fair number of PHDs so we imagine it would be an adequate culture fit.

Viacom

Additionally, if Google acquired Akamai then Viacom would be in a precarious position. The company that Viacom is suing over distribution on YouTube would suddenly be distributing their content (Viacom is a large customer of Akamai  – get off my CDN, Sumner Redstone :-)).

Rumors

There have been some recent rumors about Akamai selling so we’re interested so see what unfolds. The real question as Google discontinues its free dinner offering is:  can they spend the $5 billion to acquire Akamai?

See Also:

• Google and The Sushi Chefs

For over five years, Skype has flown under the radar with their P2P based messaging system. Because of the distributed nature of P2P, they’ve been able to deploy a minimal amount of infrastructure to support their always-on solution. Unfortunately, we think that Skype/eBay is in for a rude awakening.

As last-mile providers move to a bytes transferred model, we’re guessing that they won’t charge their customers for the bytes transferred for Internet calls if you’re using their VOIP service. This means that the once-free Skype will now start costing users a quantifiable fee.

Additionally, as computer services move away from the desktop and on to the server, users will be forced to purchase a separate piece of hardware to use Skype (the same hardware that other VOIP providers include with their services).

Overall, we don’t think Skype will disappear; we simply think this seemingly pirate business is in for some drastic changes.

Note: We at Deft Labs are not fans of a bytes transferred model for personal/home communication. We think it will stifle the communication revelation we’re experiencing. Oh, outside of a truly awful experience we had when purchasing a phone from Skype, we love the product.

According to FeedBurner, everyone has unsubscribed (including everyone at Deft Labs)! We’re sure this is simply a glitch but it’s not the first time we’ve seen major errors with the service. We’re grateful that FeedBurner is there to provide feed syndication (and you can’t beat the price – free), but we’re a bit concerned about their quality of the service right now.

For those who don’t know, FeedBurner was acquired by Google in May of 2007 for $100 million.

Update: Two hours later, it appears as if they’ve fixed the problem :-)

photo credit: pittaya

From a practical business standpoint, Kevin may be right. Based on the recent loses, Google may face an investor coup d’etat if they don’t reduce their expenses. Below is Google’s six month chart:

Landing Where?

If Google does reduce their headcount, where will the engineers go? The majority of Google’s infrastructure is proprietary, so a lot of the tools/libraries that engineers have been exposed to won’t be too practical in the market (what do you mean you don’t have Bigtable?). Of course you can argue that a talented engineer is useful anywhere, however experience with software/technology that is in widespread distribution might be more valuable to some hiring managers.

Recreating the culture at Google will also be difficult. A company like Google, which has amassed so much wealth in such a short period of time, can afford to be more liberal in their employee benefits and productivity expectations.

We imagine a lot of cashed-out Googlers will end up hiring engineers for startups.

The Chefs

Having enjoyed the luxurious Google cafe several times over the last year, we decided that the litmus test for the wellbeing of the entire company is the sushi chefs. The state of the union for the 16,805 person company worth somewhere in the neighborhood of $135.86 billion is directly related to the employment of the sushi chefs at the various campuses. Having witnessed several less than stellar quarters at DoubleClick, we understand the difficulty that must come with justifying sushi chefs to ravenous investors.

A personal note to the sushi chefs: We like cream cheese with our asparagus roles. It’s a cheesy lowbrow taste acquired at an inexpensive Sushi restaurant on Khao San Road in Bangkok, but we just can’t shake it. But, in all honesty… we’re pulling for you!

photo credit: rick

After the recent Amazon S3 outage we decided to investigate why people use S3 to serve content. On the surface it seems like a good deal but if you actually investigate the bandwidth costs, geographic locations, reporting and fault-tolerance, you’ll see it may not be the best deal out there for serving content.

Billing Models

If your application constantly serves 1 Mbps then you will consume ~ 321 GB of bandwidth per month. This of course does not include the inbound traffic which Amazon also charges to you. In reality, applications don’t tend to serve a consistent amount of data. Bandwidth consumption fluctuates up and down, the peak to mean ratio for a site is usually around 2:1. One important aspect of paying by Mbps is that you typically enjoy billing on the 95th percentile. This means that each month you can throw out the top 5% of bandwidth consumption. With a bytes transferred model you pay for everything (in and out). As network systems become more efficient, it is likely that everything will be billed on a bytes transferred model but, for the next few years, take advantage!

Costs

Now, let’s look at the costs associated with serving content in S3. Each month, you’re paying approximately $58 / Mb. This is a decent price per Mb but it’s not the best out there. Once you reach S3′s top tier you’re paying around $41 / Mb. Of course to get to this level you must be serving 50+ TB of data or a consistent 160 Mbps. The real problem is that once you’re in the top tier, you’re able to buy the bandwidth yourself for a lot less. If you use a modern CDN then you’re going to pay somewhere in the same price range as S3 but you get geographic distribution, advanced reporting and fault-tolerance as well.

Content Delivery Network

Amazon really isn’t a CDN. They serve content out of a couple geographic locations. This may work well if you’re physically near one of the data centers but for the rest of the world, it’s not adequate. Also, as the recent outage confirmed, they’re not able to distribute load in the event of a failure. Oh, it looks like Google isn’t the only Internet company using a two level name system to host content. It appears as if Amazon may be using this approach as well.

Review

We think S3 is a good system for archiving data. Amazon provides a trusted remote data center for you to store your content at a reasonable price. Would we use it to serve content real-time? No. Would we use S3 as an origin server for a CDN? Yes!

Disclaimer

The views expressed here are opinions and guesstimates for research purposes. There are insights that can be made about Internet systems; however, without inside information, it is impossible to determine the exact implementation of a system. Our goal in this post is to research the litigious nature of Akamai.

The House that ’703 Built

The lawsuit filed by Akamai against Limelight Networks is heating up. There was a summary judgment of noninfringement in favor of Limelight on patent 6,553,413. Now, Limelight has to defend their position on patent 6,108,703.

It appears as if patent ’703 may be built on a house of cards. The inventors meticulously documented a hosting system based on a two-level name system. In reading the patent, it looks like claim after claim builds upon and extends this core idea. Akamai probably used this approach because with the technologies available at the time, it made the most sense. It would be silly for a modern hosting system to use this approach. It only adds latency to the process. Taking the second level name system out of the equation would result in a drastically different patent. The only company we’ve seen using a two-level name system these days is Google :-)

SCO Who?

We’re not a huge Akamai fan. You could argue that it is Akamai’s numerous lawsuits that have impeded the distribution of content on the Internet. Even if Limelight is able to defend their system, they are still incurring the cost of a large legal battle. This legal battle doesn’t seem to have impacted Akamai. Today they announced that they had beat the Street’s estimate and growth was up 46% year over year.

Location, Location, Location

One important aspect to consider is that the trial is being held in Massachusetts, the home of Akamai and more importantly, MIT (the patent owner).

In what will undoubtedly be the largest tech deal of 2008, Microsoft has announced that they would like to get closer to Yahoo!, a lot closer. Microsoft is in acquisition talks with Yahoo! for the mind-blowing sum of $ 44.6 billion.

Forced Hand

Yahoo! has been under intense investor pressure for the last couple of years. The stock skyrocketed after the acquisition talks with Microsoft were announced but, prior to that, it had been two years of gradual decline while their main competitor’s stock was climbing.

In an attempt to regain investor confidence, Yahoo! finally parted ways with Terry Semel and brought Jerry Yang out of his semi-retired state to lead the company. It is our belief that it was Semel’s excursion into content that lead to the decline of Yahoo! While Semel was spending his time trying to do content deals he watched Google ramp up and take over. When Semel first joined Yahoo! he agreed to acquire Google for $1 billion only to have the offer later rejected by Google’s founders.

Culture Issues

Microsoft and Yahoo! probably have fairly similar cultures. Both are going to have the lifers around and thick layers of bureaucracy. Both Microsoft and Yahoo! are probably overloaded by process to ensure nothing moves too quickly.

Technical Systems

There can be only one! For this deal to make sense they will need to replace redundant systems. We believe Yahoo! will win in mail and Microsoft will undoubtedly win in search. If anyone is looking for a few thousands machines with Hadoop installed, there is probably a team at Yahoo! who can help :-) We have no doubt that the majority of ad serving will be migrated to the aQuantive systems.

Vendors

If Microsoft acquires Yahoo! then Akamai is going to be in an interesting position. We’re guessing that Yahoo! is by far Akamai’s largest customer so Microsoft is going to be in a very strong position to negotiate with Akamai if they decide to move forward with the acquisition of Limelight.

Impact on Microsoft

It will take Microsoft years to fully absorb and integrate Yahoo! While Microsoft is busy with this massive deal, Google will only continue to move forward.

Blocked

Google is assembling a roadblock to try and block the deal on the claims of an anti-trust position. If they’re successful, Yahoo! will be left dazed, confused and with a plummeting stock. If the deal is delayed then Microsoft will more than likely sit on their hands until it goes through. Google is in a slightly weaker position because it sounds like they’re reversing the position they just defended in the DoubleClick deal. Microsoft is likely to use the loss of the DoubleClick deal as a win here. Now that’s what we call reusable code :-)

Overall Sentiment

Outside of the sheer size, there is nothing inspiring about this deal. It boils down to a technology giant moving in and devouring a former Internet luminary.

Good luck you Yahooligans!

If you look at the enterprise space, IBM, Oracle and Sun have a relatively full software stack. Additionally, IBM and now Sun have a mostly complete hardware stack as well. It seems natural that Oracle would’ve liked to acquire the product (to kill), but we can’t imagine the MySQL team would ever sell to them. However, if Sun buys MySQL and then Oracle turns around and purchases Sun in a year or so, then you’re going to see Oracle sitting a lot more securely in the enterprise space. When pitching large customers Oracle always has to partner with a hardware provider in order to compete against IBM. With Sun, Oracle could completely cut HP out of the picture and capture all the revenue soup-to-nuts. If Oracle has a full hardware and software stack, HP is left selling the printers.

It is possible that Microsoft is primarily interested in hiring Limelight’s 1,000 employees that are trained and working in the CDN industry.

We believe that Microsoft would get a lot more bang for their buck if they purchased one of the smaller, more agile CDN players like Panther Express.

Note: As a co-founder of Panther Express, I may be a bit biased.

Social Networks

We don’t believe there will be the massive rush to develop applications for LinkedIn. We think recruiting, consulting and contact management applications will have a lot of success on LinkedIn. We also think LinkedIn should purchase Meetup.

Pulse is a solid product and we believe Plaxo will continue to grow this year. We don’t expect them to overpower LinkedIn but solid growth is expected. We’ll see how they do in the 3rd party application space.

With a war chest of approximately $300M (minus whatever people took off the table) we expect Facebook will continue to invest heavily in infrastructure and R&D. We expect to see their user base go to at least 100M this year. Application development will continue at an amazing rate but we’re fearful we may start to see application neglect work its way into the system. If 3rd party application developers don’t maintain their products then you will see start seeing errors. We also predict that 2008 will be a year of consolidation for 3rd party applications. Application prices are low and someone will come in and buy a lot of them.

We believe the MySpace network will continue to be cannibalized by Facebook and other social networks. They will continue to produce innovative ways to promote content but we see a lot of their users migrating to other platforms.

Valuations of social networks will come down to something reasonable and Ning will scoop up a handful of smaller players to create a large network.

Technologies

OpenSocial will continue to grow and mature. Facebook has about a 9-12 month technology lead on Google but if platforms continue to adopt the technology it should do well in the long run. OpenSocial has a long way to go to capture developer mindshare and releasing a product prematurely doesn’t help.

This is the year of OpenId! There are over 120M OpenIds in existence and the list of major tech companies that have already or are in the process of adopting the technology is impressive. Expect major growth in this area.

Browser based widgets will continue to grow strongly and monetizing widgets with advertising will become prominent.

The massive adoption of RSS has been amazing; however, too many people still don’t get it. Expect to see companies start further monetizing their RSS feeds. We don’t think the subscription model is the proper way to monetize RSS feeds. Innovative advertising will win in this market.

Blogs

Blog adoption and growth is dependent on the tools available. We expect WordPress to take a large step forward this year at the expense of Blogger and LiveJournal.

Tumblr has drastically simplified blogging. Their micro-blogging approach has gained a very loyal audience. We expect great things from this young company.

Twitter has revolutionized how people communicate. The use of Twitter in business isn’t widespread outside of the tech community but we predict its adoption will continue to grow. Truly private Twitter groups would drastically speed the adoption rate by teams.

Instant Messaging

This is more of a plea than a prediction. Firefox – Please build a universal instant messaging client into the browser. Use the Adium source if they have a compatible license.

Have a great year!

With a steady flow of new applications being released for Facebook, the platform continues to prove an excellent opportunity for everyone.

Rumors and Theories

VentureBeat believes that News Corp is interested in LinkedIn to bolster the declining Wall Street Journal classifieds business. We think this might be a nice enhancement for WSJ but that other News Corp properties stand to gain more from the acquisition, namely MySpace. We think the WSJ integration will probably work its way around but imagine that MySpace will be the number one priority. We see the stagnant/declining MySpace traffic as a much larger leak in the dam.

Read/WriteWeb believes that an acquisition by News Corp would increase the pressure on Facebook to join OpenSocial. If both MySpace and LinkedIn hadn’t already joined the OpenSocial network, we would give more credibility to this statement.

What is Murdoch thinking?

MySpace has announced that they will release a friend feed, privacy controls, application platform and an ad network. The beauty of these announcements is that they all came after Facebook already released products in these categories.

Outside of content distribution it appears as if the innovation canister has run dry at MySpace. We think Murdoch needs to clean house at MySpace and bring in people who can give the property a technology advantage. Perhaps this is the real reason behind the acquisition of LinkedIn.

LinkedIn’s Pipeline

LinkedIn must have something special in their product pipeline if they’re discussing a sale. LinkedIn has stated that they have been profitable since March of 2006 yet they took $12.8M in venture capital in January of 2007. What are they spending the money on? Take a look at our previous post that estimates Facebook’s platform costs.

If LinkedIn doesn’t have a stellar product in their pipeline they may be spooked by their drastic decline in traffic last month.

12-06-2007 UPDATE: It looks like the initial Compete numbers may have been off. Originally, they showed a 31% decline in LinkedIn’s traffic in Oct of 2007. Hmm…

What about Google?

Google is trying to sit on top of all social networks so this really doesn’t impact them. Google is already in bed with News Corp so they’ll only have to worry about fewer OpenSocial partners to integrate with.

LinkedIn Valuation

We’ve heard a couple of interesting numbers about what people think LinkedIn is worth. We’re going to assemble a valuation based on publicly available data and a couple of assumptions.

Users: 16M
2007 Growth: 264%
Oct 2007 User Visit Growth: -31%
2006 Revenue: $10M
2007 Revenue: $25M (guesstimate based on 2007 user growth)
Monthly Visitors: 6,631,828
Monthly Page Views: 74,939,656

Based on their traffic and revenue it’s hard to imagine that LinkedIn would fetch that high of a price. LinkedIn’s most valuable asset is actually your data or their social network. If LinkedIn is able to fetch $300 per user then they are looking at a $4.8B valuation.

The main problem with LinkedIn’s traffic is that their users are not visiting the site much. Last month only 1.9M or 12.3% of their users visited the site. If you compare this to Facebook’s 50% of total users visiting the site each month you can see that $300 is way out of the park for LinkedIn. We think that the LinkedIn users are worth more along the lines of $35 each. This puts the valuation of LinkedIn somewhere around $560M. At this price LinkedIn would sell at an estimated revenue multiplier of 22.

Of course you can’t discount social network hype. They may very well get an astronomical sum. Facebook was able to raise capital on a revenue multiplier of 115. If LinkedIn is able to sell for $560M then the founders, investors and employees should all be really happy.

Page Views
This post is based on 1.5 billion 3rd party application page views per month (via Compete).

Bandwidth
This is tough to estimate so we looked at the global distribution of their user base, estimated a peak-to-mean ratio based on the global distribution and then investigated the average size of 3rd party application pages. After all this research, our guesstimate is that they’re consuming a peak of approximately 1 Gbps for 3rd party applications (static video and image hosting not included in estimates). We’re going to take a stab in the dark and estimate that they’re paying $15 / Mb for this large commit and this brings us to an annual bandwidth bill of $184,320. Once again, this does NOT include any content delivery, this is merely an estimate of the 3rd party application bandwidth.

Hardware
The Facebook platform requires a lot of HTTP connections to and from external application servers so this reduces the overall number of requests per second each server can support. Additionally, to make an application like this scale you must keep most of your data in core. We’re going to make the assumption that each user profile has about 500 KB of data. If we assume each server has 4 GB of memory they would need ~ 5,000 servers to cache the 19 TB of profile data. Assuming they’re using commodity hardware, this is roughly $5,000,000 worth of hardware. We’re not going to discuss the hardware amortization in this post. Fortunately, these servers can also serve requests/content.

Hosting
If we assume Facebook has approximately 5,000 servers and each rack has enough power for 30 commodity servers, then Facebook would need an estimated 167 racks. Let’s say they’re paying $500 per rack per month (including power costs); if so, they would see a bill from Level 3 for ~ $1,000,000 per year to host their servers (bandwidth excluded).

Staff
People are usually the most substantial cost for a company and Facebook is no exception. For this post we’re going to estimate Facebook has one hundred people dedicated to their platform (engineers, sys admins, managers, marketing, etc.). Salaries are hard to guess but we’re going to say the average salary is $80,000 per year. Each employee has a lot of loaded overhead as well. In this post we’ll assume Facebook has $100,000 per employee per year in loaded costs (medical benefits, hr, office space, training, vacation, etc.). This brings the average cost per employee to approximately $180,000 or $18,000,000 per year for the platform. We didn’t consider the value of the option grants, but we’re sure they are significant.

Year One Costs
The purpose of this post is to try an estimate what Facebook’s platform costs are for one year. Since we don’t have any inside information from Faceboook, our year one guesstimate is:

Bandwidth:       184,320
Hardware:      5,000,000
Hosting:       1,000,000
Staff:        18,000,000
------------------------
             $24,184,320

Advertising
Facebook places an ad banner on most application pages under the left navigation. Looking at the numbers above, it’s important for them to monetize this space in order to keep the platform free for developers. Since Facebook is able to derive ad revenue from 3rd party applications there is nothing that leads us to believe they will start charging applications for access; however, we do think that they will eventually start charging for promotion in their application directory.

Despite the tone of previous posts we still think OpenSocial is good for the social application industry. However, we still think the product has a ways to go.

Did someone just disclose that they don’t have a competing product in the pipeline? Is this a disclosure about the Facebook investment deal? The billion dollar question on everyone’s mind right now is if Facebook will announce an investment deal before the markets open tomorrow morning.

Let’s say Facebook is able to raise capital on a pre-money valuation of $15B. We’ll also assume that they have 39M users. Not including the value of their IP or employees, Facebook is worth approximately $394 per user. Believe it or not that’s not even close to the $710 per user Yahoo! paid when they acquired Broadcast.com in 1999 from the dancing wonder Mark Cuban. Inspired by the recent Compete article that discusses how much traffic third party applications contribute to Facebook we played around with the value of the social application market on Facebook.

A warning about our numbers. All of the data we reference in our calculations comes from Compete and some of the numbers are from August 2007.

If Facebook users browse a staggering average of 44 pages per visit and the site had 350M visits in August then you’re looking at roughly 15.4B page views per month. Now if we trust the Compete numbers we see that 3rd party applications are responsible for roughly 10% of the Facebook traffic. This works out to about 1.5B page views per month or 49M views per day for 3rd party social applications.

What is the value of the 3rd party social application market to Facebook? First we have to make a few assumptions. For this entry we’ll assume the average CPC is around $0.20, all pages have one skyscraper banner/flyer and the average CTR is 0.04%. Based on these assumptions and the data from Compete we can see that Facebook’s potential ad revenue is ~ $1.4M per year for the social application traffic on Facebook. The CTR needs to be dealt with because this much traffic warrants a lot more revenue. This number doesn’t include the value of the 3rd party applications or the ad revenue application providers are able to generate.

What does this mean?

The social application industry is only five months old and it’s off to a respectable start. With more players promising competitive platforms things are about to really start cooking.

Going forward?

We think eventually, the Facebook specific traffic will shrink and application traffic will dominate the site (including the Facebook owned applications). Really, how many times can you scroll through a friend list? We can also see that only a fraction of the Internet population has been fully indoctrinated into using applications on Facebook. After Facebook closes this round we expect to see them launch a major advertising campaign. We also think we’ll see application providers start playing around with subscription models on social networks. If there is a social application you use all the time wouldn’t you pay $10 a year to avoid the ads? This would be incredibly easy if there was a Facebook payment system ;-)

The same logic holds true for social platform applications. From a technical standpoint, the anatomy of a Facebook platform request is:

1. User makes a web request to Facebook
2. Facebook inspects the request and then issues a proxy request to your servers
3. Your application processes the request and sends the response back to the Facebook servers
4. The Facebook servers process your response and displays the page to the user
5. The user interacts with your application and the cycle is repeated

OK, now that we have the two second web service tutorial out of the way we can discuss the importance to social application service providers.

If your application servers are located in New York and the Facebook servers are located in California you’re adding additional latency to your page load times.

If we were cruising along perfectly at the speed of light we would see an additional 13 milliseconds of latency for each packet coming and going. Round trip that would be about 26 milliseconds per packet. In reality, when moving on the same ISP backbone (Level 3) we see roughly 35 milliseconds latency per packet or ~70 milliseconds round trip.

Our basic testing indicates that our applications which are hosted on servers in NYC have about a 200 millisecond network tax per page. This isn’t terrible but the geographic distance and network diversity also increases the likelihood of a dropped packet, which would kill your page load time. If you’re not concerned about slow page loads in your Facebook application, you should be. Facebook will timeout the request from their servers to yours if you take too long. The result is an ugly error page.

How is this problem solved?

Social Platforms need to be geographically distributed. Facebook has enlisted the help of Akamai but that doesn’t solve the problem that all their proxy requests are coming from the same location (California).

Google should be well positioned because they already have a massively distributed system. Something only the Googlers know is if their system is capable of handling the frequent social application changes? A basic way of looking at this is to ask how long it will take for the changes in my profile to be seen by users around the world.

How is this problem solved (today)?

Move your servers to California and make sure they’re sitting in a Level 3 data center. You’ll be in close proximity to LinkedIn, Facebook, hi5 and MySpace.

If you work in marketing at Level 3 you can make the check payable to Deft Labs Inc.

Great, what does this have to do with social applications?

There is a new operating system in town and it’s not a Uni*x variant. Social Platforms are now functioning a lot like large, distributed operating systems. Google has been rapidly working towards this and you could actually say they have a distributed OS. However, I would argue that their OS is missing a GUI (apart from Google created applications). What does this mean for your laptop? You had better really enjoy your local OS because each day there are fewer and fewer reasons to write for the desktop.

Now… Do I pre-order Leopard? I’m excited about virtual desktops or in Apple speak, Spaces.

So far, social application service providers have been able to quickly monetize the Facebook platform and this will remain an important factor in driving developer support. Tools and libraries will also be essential. Facebook started this by introducing the Facebook Markup Language or FBML. FBML provides XML tags that enable application developers to embed UI functionality in their application that is consistent with the Facebook UI. This can be compared to the tools available in Visual Basic, PowerBuilder and other similar technologies. By providing developers with tools and libraries that allow them to efficiently work on business logic you reduce the time to market and the technical aptitude required to build the application.

Portability is the one caveat of platform specific libraries; although, we don’t see this as a difficult technical problem. The smart ones will embed their business/application logic in reusable components or services so they only have to port pieces of the UI.

Just about everyone has seen this but for those who haven’t witnessed Steve Balmer having a Howard Dean moment, enjoy.