The Pubmatic study that Jason writes about (coming via Mark Dempster, the Marketing Partner at Sequioa Capital) shows CPMs, the cost per thousand views of an ad, falling for social networking sites - look at the numbers for large sites in the figure below:

As Jason says, “social networking sites suck for advertising”!    However, while it’s undeniable that social networking sites are not currently great places for advertising, that doesn’t mean it has to be that way forever.

When people think about Internet ads, they tend to think of Google-style adwords ads. Without doubt, search is the killer application for these kind of ads; and to be honest, it may well turn out that search sites are the most valuable advertising platforms of all.  For the current context, though, I’m not sure that matters: the problem with social networking sites isn’t to make them more valuable than search sites.  Rather, I think it’s to make them simply highly valuable. Nevertheless, it’s clearly going to need some creative thinking to make real progress with social networking ads.

So, what’s the answer? Stepping back, the big picture, of course, is that ads need to be targeted to be valuable.  However, simply mimicking Google-style targeting may well be the wrong way to go.  I suspect it may turn out that social networks need not only a different way of targeting ads, but a different kind of advert altogether.

Now, as Jason rightly points out, when people are on a social networking site, they may not be thinking about making a purchase.   What does that mean for the ad strategy?  Simple - don’t run straightforward ads that simply expect a click-through for a near-term purchase.   The obvious tactic would be to use brand advertising.   However,  that doesn’t really address the fact that targeting adds hugely to the value of the ad.

My suggestion is as follows. Create a new style of ad, based around special offers and promotions, which in turn would be based on people’s interests as expressed by the social networking site they’re on and the individual profiles.   People are susceptible to special offers and promotions, even when they’re not actively looking to buy something.  Why?  Easy - free stuff and genuine money off offers get people’s attention.  For example, imagine you like mixing drinks, and have listed “cocktails” as an interest on your social networking profile… Then, if you see a “Buy one bottle of premium vodka, get a free sample bottle of our new pear flavour vodka free!” that might get your attention, even if you hadn’t previously been actively thinking about a buying another bottle of vodka.  Or, if you like playing golf, and you see an offer saying, “Want to play to the new competion rules? $80 dollars off our new driver, if you buy before the end of the month.”

The value to the consumer is clear - they can get some great deals on stuff they’re interested in. There are benefits for the advertisers, too. Companies often have promotions and special offers on their goods and services.   The benefit of giving companies the chance to target these offers, is the opportunity to connect their special offers with customers who are not likely to simply take the free stuff and run; but instead to attract good return customers, who will buy more of their products in the future.

By the way, I should say that this isn’t a new idea.  Credit companies have been doing this kind of targeted advertising for years; offering promotions and offers to customers based on their previous purchases, which of course they know all about.   It works for them, and it might just work for social networking sites too…

Unfortunately, I think Powerset has two major problems: firstly, the kind of approach Powerset appears to be taking performs worse than keyword search; and secondly, Google is, in a low-key way, already implementing a “natural language” approach to search that does work better than keyword search.

Back in the late 1990s, I developed a natural language search system, based on a model of how the human brain understands language, for use in the biotechnology and pharmaceutical industries (you can read a bit about it in this publication on the American Chemical Society web-site - it’s an easy read, with no overly technical content in it!).     The key thing that made our system work, was that we built in detailed knowledge about the particular domain we were interested in, and the system was designed from the ground up to make use of domain knowledge.    General purpose, or domain-neutral, approaches to natural language understanding software, by themselves, simply don’t work very well compared to keyword search.

What does this mean in practice? Well, when you type in your natural language query into Powerset, it means you’re unlikely to get a clear answer to your question, because Powerset’s domain-neutral software doesn’t really understand the question.   However, if you type in a natural language query into Google, and it’s specific type of question that the Google software understands, then you get a great answer back.    Here’s an example query -  “What is Elton John’s real name?”

Powerset Results

On Powerset, the results are as follows:

1. Elton John

   (A spoonerism for “Budget Rent-O-Van” and a reference to his real name Reg.) … Spots related to Elton John’s childhood and career

2. close

   Elton John discography

   Elton John’s Milestones (1980) … | 1994 | “Ain’t Nothing Like the Real Thing” (with Marcella Detroit) (Nickolas Ashford/Valerie Simpson) | Duets | 24 | (N/R) | (N/R) | – | – |

3. close

   Elton John’s Greatest Hits

   Though the album includes many hits from Elton John’s early era, it is noted for leaving off several of his more popular songs.

4. close

   List of albums (E)

   Elton John – Elton John’s Greatest Hits Volume II (1977) … Evermore – Real Life (2006)

5. close

   Elton John’s Greatest Hits Volume II

   Elton John chronology Blue Moves (1976) Elton John’s Greatest Hits, Vol. 2 (1977) A Single Man (1978)

Not exactly what someone asking that question is looking for.  It’s worse than a keyword search.

Google Results

On the other hand, if you type the same question into Google, you get the following answer:

Elton John — Birth Name: Reginald Kenneth Dwight

According to http://www.movietome.com/people/3666/elton-john/index.html - More sources »

That’s a perfect answer.  Google’s software understood what I was asking, and gave me a clear answer back.    How did it understand the question?  It didn’t take a domain-neutral “subject verb object” style approach to trying to understand the question.  Instead, the software “knows” that a search with “the name of a person” and the phrase “real name” is most likely to mean, “What is this person’s birth name?”.

I really think Powerset has a lot of work to do if it wants to be taken seriously as a player in search.   The truth is, as of now, Google gives better results for most searches, including some natural language searches.    That’s not to say Powerset can’t find a niche where it can succeed - but, in my opinion, promoting general natural language queries isn’t the right way forward for the company.

If the VCs behind these funds are serious about investing this money in these areas (which they may, or may not be), that means we should expect, over the next few years, that five to ten really significant, well-funded, new start-up companies will emerge with a strong focus on iPhone or Blackberry.   What will these companies (need to) look like? Well, to count as a success from the point of view of their investors, each of these companies will need to achieve valuations of at least $200M within five years of being funded.

If such a valuation were to be based on revenues, that means the company would need annual revenues approaching $30M.   Let’s assume that in five years, there are 100M iPhones or 100M Blackberries in the market.    What if the company achieved 100% market penetration i.e. had 100 million users? That would mean the company received $0.3 per year from each user of the device.   With 10% share i.e. 10 million users, then they would need  $3 per year per user.  With a 1% share i.e. 1 million users, then they’d need $30 per year, per user, and with 100,000 users, then they’d need $300 per year per user.

A target revenue of between $30 and $300 per user per year could be realistic, depending the type of application or servce, and the target market. So that means opportunities where there’s a good chance of attracting between one hundred thousand and one million paying users could be interesting to the investors behind these funds.

If the valuation were to be based on numbers of users in the short-term, rather than revenues from users, then it would need to be an opportunity where there was a realistic route to achieving millions of users.   There are some obvious opportunities there: mobile search, mobile social networking etc.  It would be surprising if at least one mobile search and mobile social networking play didn’t get money from these funds.

All the opportunities, though, probably need Apple and RIM to deliver significant growth in device sales, from where they are now.    There’s a risk they that they won’t - smart phones have all but been failing in the market for the last five years. That means there’s could be big competition for developer mind-share from the massively larger feature-phone market (there are already billions of these devices in the market) providing that the technical problems with developing for feature phones can be resolved.

The problem with the feature phone market has been device fragmentation:  that is, you can’t develop an app, and have it “just work” on billions of different models and makes of phone.  Last week, though, Sun Microsystems threw a spanner in the works.  The company announced all-new, and surprising plans for JavaFX Mobile - the aim is to make JavaFX mobile apps run near-identifcally on every device produced by just about every handset manufacturer .    It’s a highly ambitious plan… but what if it works?  That would will be truly transformational for mobile application development because it would mean a mobile application or service could be used by billions of people.  You have only to look at the single current example of such a service - SMS text messaging - to see just how massively valueable that could be.

If you want to build a venture capital-backed mobile application/service business based on revenues, then building for iPhone and/or Blackberry is a reasonable bet, assuming the market increases in size rapidly from where it is now.   People that buy iPhones and Blackberries can at least afford to buy applications or pay to subscribe to services; and there’s a good chance they’ll be using these phones for business; so the Enterprise opportunities could be significant.   If, on the other hand, you want to build a free app or service, with really huge numbers of users, then JavaFX Mobile (which will likely run on Blackberries, but perhaps not on iPhones) could very well turn out to be the best platform to develop for.

I’ve been aware of JavaFX, Sun’s new Rich Internet Application (RIA) platform, since it was first announced a year ago at JavaOne 2007 (and before that, even, when the JavaFX Script language was called F3).   A year later, the demos at JavaOne 2008 are looking pretty interesting; and the new cloud systems and services, built to interact with JavaFX, that  Sun announced a few days ago (codenamed Project Hydrazine and Project Insight) are intriguing.  So far, then, so good.

I watched Dan Farber’s interview with Sun CEO, Jonathan Schwartz, over on CNET News.  The strategy sounds as great as the technology.  Jonathan said (9 minutes and 3 seconds into the interview),

Free and open are the two attributes that ensure access to 100% of the market place…  That applies to RIA platforms as well.  So, our ability to say faithfully, and with integrity: we have free platforms that are Open Source… lends credibility and drives volume.

I tend to agree. From this, then, I took it that JavaFX was to be a free and open source RIA platform.   I could be wrong - I may have misunderstood what he was saying - but I’m pretty sure that was the message Jonathan intended people to take away.

Overall, then, JavaFX sounds like a great technology. In fact, it was all looking interesting enough for me to take a serious look at starting to use JavaFX, to begin to evaluate it as a technology for our future real-world use.  I headed over to Sun’s JavaFX web-site and thence to the JavaFX FAQs.  I have to confess - I was more than a little surprised by what I read. Here’s one of the questions, and the answer:

Is JavaFX technology open source?

JavaFX technology is very early in its development. The JavaFX Script language, currently being developed with the community’s help (see OpenJFX project), will have a grammar and syntax that are open source. Some parts of the language are already open source. The JavaFX compiler, runtime engine, player, and tools currently under development are not expected to be open source. You can participate in the OpenJFX Compiler Project, which focuses on creating a JavaFX compiler to translate JavaFX scripts into JVM class files (bytecode). This compiler will leverage and extend the JDK’s javac compiler capabilities.

Update 14 May 2008:  the text of the FAQ has been ammended, to read…

We continually solicit and get feedback from the open source and Java developer communities about key elements of our technology, and this ongoing conversation often influences our product development.

To this end, several components of the JavaFX product family are already projects in open source, including JavaFX Script, JavaFX Compiler, JavaFX Script Plugin for NetBeans, and Scene Graph. See the OpenJFX community site for information on how to get involved and contribute. Sun will continue to engage the OpenJFX community as we release JavaFX products. This fall we will be rolling out our open source strategy for JavaFX technology concurrent with the release of version 1 of JavaFX Desktop. In the meantime, developers can go to the OpenJFX community site for information on all the technology components that are currently available.

So, that’s a “no” then! Let me pick out a key part of this answer… The JavaFX compiler, runtime engine, player, and tools currently under development are NOT expected to be open source (capitalization is mine).  In other words, JavaFX is mostly going to be closed and non-free.

Something doesn’t add up - compare the Sun CEO’s comments with what’s written on the web-site  (by the way, in case anyone at Sun is in doubt, and/or has a different opinion, what the CEO is saying is the right strategy). At the very least, the messages around JavaFX are unclear.   I’m pretty sure that potential JavaFX developers would be interested in getting some better understanding on this.   Certainly, the lack of clarity has stopped me getting my hands dirty with JavaFX technology for the time being.

For me, it’s not a matter of whether it’s open or closed… it’s simply a matter of wanting to know what it is, one way or another.  Of course, the answer might affect my choice (as Jonathan says, free and open guarantee access to 100% of the market), but the point is this - not knowing the licensing terms of a technology is just about as big a barrier to adoption by software companies as you could ever imagine…

JavaFX Applications Will Run Everywhere

The big surprise of the keynote is what Sun has been doing with JavaFX.  It turns out that JavaFX applications are going to run everywhere, and in some pretty interesting ways (including using audio and video).  Many people had been assuming that JavaFX Mobile was going to be a small niche play… and that, perhaps, no handset developers would be interested in signing up.   It turns out that Sun has been able to make JavaFX applications run not just on Java SE, but also on Java ME (and they also demo’ed a JavaFX app running on Google’s Android).

This is a game changing development.   It means that JavaFX applications can run on any device that runs Java, even low-end feature phones.   Pretty much every handset manufacturer appears to have signed up (we didn’t get details on this, though, so I might be wrong).   JavaFX represents an opportunity to let the same application run consistently across billions of screens… which is a huge deal.

The way that JavaFX works on the desktop looks pretty interesting too.  A JavaFX applet running in a web browser can be “torn” out of a web page, and allowed to run as a standalone desktop application with a single drag and drop gesture.

If Sun delivers on what they’re promising, it’s really going to take Java-powered devices and applications to another level.

A couple of new projects were also announced at the Keynote, but were glossed over.  They’re to do with how the back-end of these new types of application, that can run consistently on billions of devices, can be handled. The first of these is called Project Hydrazine - an all-new cloud computing platform (maybe it’s the answer to the question I posed on this blog yesterday - what’s next for Network.com? ); and the second is Project Insight - which is a new information-gathering platform that could enable targeted advertising across the billions of devices that are running JavaFX applications.

All-in-all, an interesting keynote.   Because nothing is shipping yet, though, everything is subject to change before it hits the market… and, because the devil is in the detail with this kind of thing, we won’t know how well everything really works until the technology ships.

Update: The JavaFX.com web-site is now live.

Today, we’ve had the first glimpse into Sun’s future plans for cloud computing, which will be based around the brand new OpenSolaris OS (just released today, after being in preview for some time).  You may, or may not, know that Sun already has a cloud computing offering, called Network.com.  So far, this has been focused on executing batch parallel computing jobs on a pay-as-you-go basis.  That’s fine, for what it is.  However, batch computing jobs are a small niche compared to running a “complete” pay-as-you-go cloud computing offering, with scalable clusters of web servers, middleware servers, database servers, and scalable reliable back-end storage etc.  That is, an offering that would let you get up and running building the next Google, Facebook or YouTube…. all with no upfront investment in hardware required.   So, what are Sun’s future plans for cloud computing?

For sure, I think Sun is going to re-invent Network.com in the future… but not today.  Today, Sun announced the kind of complete offering I described above; not as its own offering on Network.com, but by partnering with Amazon to deliver OpenSolaris as an optional operating system on Amazon’s super-popular Elastic Compute Cloud (Amazone EC2). It goes into private (by invitation) beta today.

Amazon EC2 (combined with Amazon S3 storage) is an interesting option (perhaps the best option currently around) for people and companies deploying systems that they hope and expect to enjoy rapid growth.   Choosing OpenSolaris as the OS on EC2 will be a no-cost option.

This is an intriguing move on Sun’s part, and one that begs the question… “What’s next for Network.com?”

• Cloud computing - we’ll hear exactly what they’re doing with Amazon EC2, OpenSolaris and maybe even MySQL
• Mobile - we’ll hear about JavaFX Mobile (perhaps an announcement of an SDK), and maybe what they’re doing with Google’s Android, and Apple’s iPhone)
• Java on the desktop - we’ll hear about the release of the new consumer Java which is imminent;  and maybe get the first public viewing of new, upcoming support for video in Java
• Developer tools - perhaps the first public showing of JavaFX Designer which is in development

Could be an interesting few days…

If Yahoo!’s shareholders are really holding out for an offer of $37+ per share, as some are suggesting, I suspect they’re going to blow it, and end up with nothing.   No-one else wants to buy Yahoo!   Microsoft’s is the only offer on the table.   It’s hard to see how Yahoo!’s share price would ever recover if Microsoft walks away.

And that’s just what happened.  Yahoo!’s negotiating team, presumably after consulting with shareholders, dug its heals in, and, yesterday, Microsoft walked away, withdrawing its offer. In a letter to Yahoo! CEO, Jerry Yang, Microsoft CEO Steve Ballmer wrote,

In our conversations this week, we conveyed our willingness to raise our offer to $33.00 per share, reflecting again our belief in this collective opportunity. This increase would have added approximately another $5 billion of value to your shareholders, compared to the current value of our initial offer. It also would have reflected a premium of over 70 percent compared to the price at which your stock closed on January 31. Yet it has proven insufficient, as your final position insisted on Microsoft paying yet another $5 billion or more, or at least another $4 per share above our $33.00 offer.

Yahoo!’s Board and its major shareholders appear to believe they can grow the company’s value by a huge amount in the coming years.   I haven’t done the calculations to factor in risk, but make no mistake, if you turn down an offer to pay a 70% premium for the current market value of your company, you must believe you can grow the value by hugely more than 70% over a period of a few years (perhaps by 200% - 300%).  I’ve not seen a plan from Yahoo! that can deliver this.

Now, I’ve said here that this is an outcome that Yahoo!’s shareholders would be happy with.   That is, they wanted $37 per share, or nothing at all. The big question is: will they still think this when the cold light day hits on Monday and trading of Yahoo!’s stock resumes?   How low will Yahoo!’s share price go?

If, by some chance, Yahoo!’s Board let Microsoft walk away without the approval of their major shareholders (which I find almost impossible to believe)…well, oh dear…  that will be the end of Yahoo!  The shareholders will fire the Board.  The management team will be fired by the new, replacement Board.   And Yahoo! will be sold to the highest bidder.

Update: Monday 5 May, 10:48 GMT - The Sell-Off Begins

Yahoo!’s stock is not only traded in the US… In Europe in early trading, Yahoo!’s stock was already down to $23.17 a share…

However, yesterday, Sun surprised investors by posting a quarterly loss.   To address this, Sun announced job cuts.  Yes, this will help with profitability.   However, the really big questions surround Sun’s revenues, which aren’t growing.   That’s pretty surprising, given the strength of the company’s offering compared to the competition. So, why isn’t the company growing its revenues (they seem to be staying pretty flat at between $3.2B and $3.8B per quarter)? I can’t claim that I’m 100% sure why, but I can think of a few things that might be both contributing factors and also easily and quickly addressed…

Sun Isn’t Maximising The Potential Of Its x64 Hardware - People don’t know you can buy x64 hardware from Sun, and they don’t know the servers are certified for Windows 2003 Server

I know a few CIOs, COOs and CEOs of companies that buy reasonable amounts of x64 server hardware.  They tend to buy HP or Dell.   Recently, when chatting to one of these people about a new data center they were building, I asked if they’d looked at Sun’s x64 servers.  Their reply?  “Oh, I didn’t know Sun made x64 hardware… and anyway, we want Windows 2003 Server for these, not Solaris.”

You can’t buy an x86 box from Sun with Windows 2003 Server pre-installed

Whether people like it or not,  Windows 2003 Server has a pretty big market share; I think around 60% of servers ship with Windows.  If you want to buy a server from Sun, you can have it with Solaris pre-installed, or a couple of flavours of Linux pre-installed… but if you want Windows 2003 Server, you have to get the license yourself, and install the OS yourself.  The web-site tells you, “This OS is customer-supplied”.

Now, Sun signed a deal with Microsoft in September 2007 to become a Windows OEM.  They were supposed to start selling servers with Windows pre-installed within 90 days from then.  I have no idea what happened; and haven’t seen any communuications from Sun on this matter since that announcement.

Sun Is A Systems Company, Not a Box Company - But That’s Not How Its On-Line Store Works

Sun is not a “box company”  - they spend billions on R&D each year.  They’re a “systems company”. They offer servers that are genuinely innovative and well-designed. They also have one of the most impressive (or it will be, when it’s all finished!) software stacks in the technology industry: data center management; server-side hardware virtualisation; operating system; relational database; application server; cross-platform desktop application layer;  cross-platform developer tools; and desktop OS virtualisation.    However, when you visit the on-line Sun store, it looks like Sun is a box seller not a systems seller: the workflow of the purchase is all about individual parts, not about buying systems.

For example, let’s say I want to buy the following system: an x64 server running Windows 2003 Server, with some disk-based storage, and some tape backup storage, along with the relevant software to make it all work, pre-installed.   That would mean, when the hardware arrives, all I have to do is plug everything in, boot it up, and I’m ready to start configuring the system.    Well, I can’t do that with the on-line Sun store.  I have to work out which bits are compatible and have the right features myself; and then go buy my OS from somewhere else… and then, when I come to set it all up, run the risk that I’ve bought the wrong bits and so not everything will work.   I want to buy the system from Sun in such a way that I know everything will work together (for low-end systems, it should be necessary to go through a value-added reseller).

The above was just a simple example.   Why can’t I go to the Sun store, and do something like the following workflow, which would take advantage of Sun’s unique offerings:

1. Specify the kind of server I want, and the amount of storage - e.g. two quad-core CPUs, 8GB RAM, with 1TB SATA storage on-board.  The web-site would then recommend potential servers to look at.
2. Specify that I want virtualisation software pre-installed, and OSes installed on top of that…  so I’d like both Windows 2003 Server and Solaris 10 to be preinstalled on top of a pre-installed virtualisation layer.
3. Specify that I: don’t want to pay for support of the virtualisation softare (i.e. I’ll have that for free); don’t want to pay for support of Windows 2003 Server; but that I would like to take out support for Solaris.
4. Specifiy MySQL and Glassfish app server on Solaris, and just MySQL on Windows 2003 Server
5. Specifty that I want support on Glassfish, but no support on MySQL

Now, I don’t know how many potential customers want to buy their systems this way, so I have no idea if this kind of thing would affect revenue growth.   However, I do know that Dell, for example, puts a lot more effort into its on-line store than Sun does. At the very least, putting some “systems features” into the Sun store would differentiate Sun’s on-line offering from those of the box sellers.  Also, I find it hard to believe that at least making it easier for customers to buy systems running Windows wouldn’t be good for business.