Semantics Incorporated

The Next Web to Be User-Centric (Thoughts on David Siegel's Pull Book)

Tell me, how much time do you waste searching for stuff on the web, or filling up forms with similar information over and over?

A Scattered Web of Destinations

For all its might, utility, and growth, what we have today is a scattered web, a web of destinations, on which finding information requires a whole expedition across loosely-connected archipelagos of data, each with its own information requirements*, rules of engagement, and gravitational attempts at capturing your time and money.

* I'll get sick if I have to fill one more login form. Even with the advent of Open ID and other unique identity services, way too many applications still require their own sign-in process.

So, when David Siegel contacted me to review his upcoming book Pull: The Power of the Semantic Web to Transform Your Business, I was immediately attracted by his core premise, the creation of a web that would automatically wrap around us and serve us based on the actual characteristics and needs of our lives.

"This book describes the pull era, where customers pull everything to them on demand – products, services, information, knowledge, and advice. Much of the foundation for pulling is called the semantic web, a new way of packaging information to make it much more useful and reusable. Over the next ten to twenty years, it will change business from a lead-push model to a pull-follow model of interacting with customers."

- David Siegel, Pull

Siegel describes the web not as it is, but as we'd like it to be. His vision should be a no-brainer. In the Web-powered world he describes, you would avoid repetitive information tasks and have just the information you need at your fingertip. For many of us, having the web's knowledge organize around us stands right up there with "getting more money" in our Top 5 Desires Chart.

Siegel's vision should be a no-brainer for the web authorities of this world, but it's not.

Offering The Vision We Lacked

Voluntarily or not, by escaping the "Knowledge Curse", Siegel is addressing the main shortcoming of the semantic web development efforts to this day: a vision by techies for techies. The idea of "Turning the Web into a Giant Database" powered by the semantic web stack is at the core of Tim Berners-Lee and the W3C's effort so far at formulating the next web. It's not wrong, but it's not user-centric and it remains a very incomplete vision of what is required to deliver a vastly more useful web.

To get there, the W3C approach really has been to apply the concept of Relational Database to the whole web, throwing in some innovations with RDF and OWL standards and then developing SPARQL on top (the equivalent of SQL in RDBMS). And now only, they are starting to really think about the Proof and Trust levels of the stack. This is putting the cart before the horse.

Certainly illustrations and even applications have been offered (see a recent effort by Talis with the video The Semantic Web and Linked Data – In Action), and there are benefits, but by and large those fall vastly short of what users need and what they hope for, because quite simply those benefits stem from a largely clustered and technologically-centric perspective - and from organizations like the W3C with few people truly thinking in more end user-centric terms. 

Because of this, the stack is a castle of cards: it is built on shaky foundations because little attention was paid to how the whole structure should look like and what it should do in the end. So it's overbuilt on some aspects (e.g. turning supergranular pieces of information into triples) and underthought on many others (e.g. origination, trust, speed of querying a web at large, tagging more conceptual pieces of information, timestamping etc). Seconding recent words by Alex Iskold, I will thus predict for 2010 that the "Semantic Web as described in the original vision is still not happening and Linked Data is not unfolding as quickly as some people have predicted."

It's sometimes ok to figure out some of those things as we go, but then we need at least a vision of where we want to be. As the adage says, if you don't know where you're going, any way will get you there. If the vision is to turn the web into a Relational Database as we know them, then that's what we'll get, with all the shortcomings alluded to above.

So, inspired by Siegel, I would propose that the vision and Ariadne's thread of the Semantic effort should be to enable a User-Centric Web.

Product Requirements for the User-Centric Web

To go from the web we have to the web we want, the first step is not to look at the existing technologies and wonder where to go from here, but to identify the unmet needs of end users - and by end users, I don't mean people who build the web, but people who use it for real-world purposes.

The antidote to geek's myopia is to understand the desired benefits, starting with the end in mind, and only then figure what we have and what we lack. If you do a poor job of the first phase, anything you do in the second one is unlikely to amount to much.

What's most to like about Siegel's book is that it does just that: it starts from the needs and hopes of end users, and thus can help us reverse-engineer the "product requirements" for any next-web application.

The most controversial thing about Pull is David's use of the expression Semantic Web in a very different manner from the one. To many, this will look as an attempt to surf the "Semantic wave" for the author's own benefit, and in fact I have no doubt it partially is, given that surfing on trends is pretty much the only way to sell a book concept to a publisher these days. But it is not just that.

His definition of the Semantic Web is one I have long advocated on this blog: a broad one. If it furthers the understanding of data by machine ("semantic") and it links in and out ("web"), it's part of the semantic web. Ok, for the "semantic" part, Siegel also adds a more utopian requirement that "it needs to be tagged in a royalty-free format, governed by a nonprofit organization, that all software programs can understand". We'll see whether the nonprofit part of this vision becomes true, but overall it sounds like the right direction.

Pull does an impressive job of pulling together a number of trends around one common wish. It truly is the book that was missing from our landscape, to articulate the benefits of semantic technologies to businesses and society. This is recommended reading to all builders and users of the next web, and by that I mean anyone with a web access. We have the right vision now, so let's get to work to realize it.
 

Related articles by Zemanta

• ReadWriteWeb Interview With Tim Berners-Lee, Part 1: Linked Data (readwriteweb.com)
• When Linked Data Rules Fail (mkbergman.com)
• Linked Data In Action at Online Information 2009 (blogs.talis.com)
• Technology Trends #3 - The Data Explosion (realtea.net)

Slow Adoption of Linked Data: Why?

During the November issue of Semantic Web Gang podcast, we also talked about DBpedia, which led me to share new thoughts about Linked Data. And more specifically, two things:

1. my growing concerns that adoption of Linked Data is going slower than expected (an observation I will test further) given the R&D investment and focus of a large community including the inventor of the web and the W3C, which led us to anticipate a faster growth curve. 
2. the possibility this slow adoption rate might be due to inherent flaws in the design of Linked Data. Even if adoption is in fact rapid (which I don't think it is), I am nonetheless concerned those flaws could also limit the potential of Linked Data to "turn the Web into a database", as per the main promise associated with the Semantic Web by its apostles (to this day...). 

My thinking on this topic has further progressed following good discussions with a number of people including Bill Roberts of Swirrl, Ian Davis of Talis, and Pawel Lubczonok of Thought Express (I have a business relationship with Thought Express, although the views expressed here are my own entirely, developed prior). I will expand on this in the future as I keep learning about the flaws I suspect (please share your inputs either through blog comments or by emailing me at gregboutin -AT- gmail.com).

In the meantime, I would recommend some good blog posts closely related (or even directly related, in the case of Bill's post) to all of the above:

• Delivering Linked Data quickly by Bill Roberts of Swirrl
• What’s in a name? And the Linked Data Police by Richard Cyganiak of DERI (see my comment there too)

And some other tidbits I came across on the adoption of Linked Data:

• This Semantic Web Job’s For You, by Jennifer Zaino, SemanticWeb.com Contributor - showing 3 ads only on a popular start-up job board for jobs requiring RDF skills, 15 quoting the "Semantic Web" and 3,000 for "web 3.0" (a much broader term including alternative approaches to realizing the Semantic Web - and also some complementary ones, although those would probably make a direct reference to Linked Data skill requirements)
• Observing the Growth of the LOD Cloud, by Alexandre Passant, which shows that about 55 sources have joined the Linked Data cloud in the past 12 months, and about 25 in the 12 months before that. The conclusion seems to be that there is growth. Although technically, that's true, I find it hard to see the addition of such a small number of domains to the cloud as the explosion that we were expecting, 7 long years after Tim Berners-Lee described his vision for the Semantic Web in Scientific American, and almost 2 years after he declared the Semantic Web "open for business" 
• Machine-Readable Web Still a Ways Off, an acknowledgement by Tim Berners-Lee himself of the slow adoption of Linked Data.

As some in the Linked Data community are quick to attack the reputation of commenters not sharing their enthusiasm unequivocally, let me remind those that asking questions is very much a constructive (if an unrewarding) job, helping strenghten the efforts and push them in sometimes new and better directions.

Certainly, offering solutions is important, and I try to wherever I can, but it should not be a pre-requisite to highlighting some possible flaws. If someone had done that for subprime mortgages prior to the financial crisis, we would not have gone through the mess we did. Assuming anyone listened. 

I have been an unbiased (I believe) fan of all semantic web efforts from the start of my involvement (see the early posts on this blog), but with time I have come to believe that Linked Data might have some flaws preventing it from delivering on the promised benefits. Certainly it does offer interesting possibilities and advances, but so did the Concorde!

Is Linked Data a new Concorde or Artificial Intelligence market failure in the making? I don't know and wouldn't go as far as to assert that. But I believe it's time we think deeper and take a closer look at possible limitations in the design, to ensure a positive ROI on linked data efforts.

The benefits of Linked Data have been very slow at materializing, and I'm not sure we should buy the argument that it's hinging on yet another R&D effort or the adoption of those standards by more web players. If it's going to be of any use, it should be useful today, not in another 7 years.

To be continued.

Related articles by Zemanta

• What's in a name? And the Linked Data Police (dowhatimean.net)
• Linked Data and the Semantic Web: What Are They and Should I Care? (slideshare.net)
• The end of Search? Linked Data, Semantic Web & thoughts. (webr3.org)
• Querying Linked Data with SPARQL (slideshare.net)
• Preparing Yourself for Web 3.0, LOD and 2010+ (webr3.org)
• Justifying the cloud (accmanpro.com)

Discussing the Semantic Portals of Eqentia with the Gang

Paul Miller has published the Semantic Web Gang podcast we recorded in November, in which we first talk about Eqentia with its CEO (and also friend of mine) William Mougayar.

Eqentia is a web application offering pre-made vertical portals leveraging an ontology-based search engine to extract and distribute relevant information.

As I understand it, its main differentiator with other news services is that it extracts information from the deep web based on content relevance rather than popularity metrics (unlike Google News in particular), and outputs an organized digest that offers a better signal-to-noise ratio than most news source. I have been relying on it for many months now as a user and found it very useful - Disclosure: I'm a supporter of Eqentia and became a friend of William, but have no commercial interest in the application.

Eqentia also requires the creation of ontologies by experts, although William foresees that this function and others will progressively be crowdsourced. On that note, I also recommend The Death of Taxonomies, revisited, by Theresa Regli, Principal at CMS Watch, which unlike the title suggests takes a look at how taxonomies are moving away from their rigid, hierarchical origins, to a more flexible model.

Does the Adoption Threshold of RDF Limit the Grassroot Potential of the Semantic Web?

In the October of the Semantic Web Gang podcast, Tom Tague of OpenCalais, Benjamin Nowack of Semsol, and yours truly discuss RDF and its controversial relationship with the broader concept of the Semantic Web. We tackle adoption and marketing issues facing the semantic web, and what could be done to overcome them.

A great question that applies to start-ups both in the semantic web and beyond, from Tom Tague's closing comments: are you selling solutions or technology?

I greatly enjoyed the chat and hope you will too.

Related articles by Zemanta

• The Semantic Web Gang talk about semtech2009, LIVE (cloudofdata.com)

Hybrid Approaches to Taxonomy and Folksonomy

Semantic Technology 2009: Hybrid Approaches to Taxonomy and Folksonomy

View more presentations from Earley.

Discussing government open data and data.gov with the Semantic Web Gang

Paul Miller has just put online our semantic web gang podcast for this month. We were joined by Brand Niemann of the United States’ Environmental Protection Agency (EPA) for a discussion on the efforts to apply semantic technologies to Government data in the USA and elsewhere. I also touched briefly on the Twine news at the end, and look forward to discussing that some more in the future, especially in the context of user utility, usability, and monetization models for semantic web companies.

See the following links on government open data:

• Linked Government Data (Sept. 11, 2009)
• Publishing Open Government Data (W3C Working Draft 8 September 2009)

Twine Confirms Traffic Drop on ReadWriteWeb, but Spins it Out of Control

ReadWriteWeb just wrote about my last post and relayed Twine's response, which in a nutshell is that they saw a 20-25% traffic drop but that it doesn't matter because they are all focused on version 2.0 which will free their traffic acquisition model from its dependence on Google rankings.

Couple of immediate comments (also see comments on RWW site):

Compete.com shows a 85%-90% drop, and so does Quantcast. They tend to be reliable in quantifying % changes and even Twine has endorsed them in the past. Twine has a huge incentive in downplaying this, since a drop of the scale Compete.com shows could lead to a withdrawal of investors' support. So, until I see actual proofs, my assumption is that the 20-25% drop claim is a misrepresentation.

Re: Twine 2.0 as the savior, RWW calls out the spin of the company, writing that "once again Twine is hyping itself up". According to Compete.com, 92% of visitors were passersby, meaning not repeat users. That confirms that Twine 1.0, as lots of commenters have pointed out, failed to make repeat users of the 100,000s of visitors coming to the Twine site from Google. Now that Google has pulled the plug on the bulk of that traffic (that reminds me of Geosign, which went from over $100M in revenue to zero, literally overnight a few years back - there is another analysis to conduct on that Google practice), Twine 2.0 might well achieve great conversion, but traffic-wise it will likely have to rely mainly on buzz, and much less on the search giant. Which means they are starting from close to zero, i.e. some user following minus the badwill they've accumulated from many folks like me plus now the lack of confidence in management created by the traffic drop (I'd expect a little better as an investor if I had put over $20M in the thing).

As RWW rightly points out, Twine can't afford any mistake now. And yet they are already trying to take us for a ride, again. Given that and the past inability of the company to integrate feedback and create positive buzz, I'm not holding my breath. In the meantime, Twine sure wins the title of "king of spin 2009".

Twine in Freefall?

What happened to Twine? Earlier this year, VentureBeat characterized Twine as an "early success", based on information from Compete.com. ReadWriteWeb followed suit in March explaining that Twine could soon surpass Delicious.com. In May, Techcrunch made the observation that Twine had overtaken Friendfeed in traffic. Fast Company wasn't as fast, but in June it echoed the comments with an article labelling Twine as "red-hot".

Twine indeed surpassed Delicious in May and June. That was shortlived, however. Since then, the same source VentureBeat used to discuss Twine's traffic, Compete.com, which Twine has endorsed a number of times, shows a huge drop in traffic. According to the traffic tracking application, Twine's traffic would have been divided by 10 between an April peak of 2M unique visitors and the August number of just about 208,000. Quantcast, on its end, reports peaks in April and July above 500K "people", together with a similar drop in August with 50,000 visits only. I don't know which source is correct about the volumes, or how their definitions of unique visitors and people differ, but the trend is what matters and it is striking. Alexa, generally much less reliable, doesn't show a huge drop, although it doesn't show any increase either.

According to all 3 sources, Twine is now back well below Delicious and Friendfeed. Have you read anything about that? I haven't.

So my questions: has Twine lost all his market momentum? Is Compete.com mistaken (isn't it what start-ups often claim after endorsing those sources, once south trends start showing)? Why has no press coverage been reported by Twine since May? Is it an anomaly, or the calm before the storm, and Twine is about to unveil a number of killer features and claim its groove back? (in my observation, once you lose momentum, it is very hard to get it back.) And why are none of the usual tech sources talking about it? Is it because it's riskier to report critically than to praise companies that succeed? Or has the summer just taken its toll?

Pictures speak 1,000 words:

 

 

 

This is a good illustration of the post I wrote last Friday on my other blog Growth Times (previously To Revenue, I just completed a name change), Welcoming Constructive Criticism, a Key Success Driver for Start-Ups.

For the record, back in February I published an interview with Nova Spivack, the founder of Twine, in which we took a 360-degree view of the application, with a focus on some concerns raised on a VentureBeat discussion in the past. I published the interview as is but added personal comments on usability stats provided by Twine, in my view showing what I have long suspected to be low engagement. Nova took offense at this public criticism and instead of looking at the feedback itself and address it calmly and constructively, he decided to libel me on twitter, through tweets he has since deleted. Ok, I know, I sound like a conspiracy theorist - but anyone who knows me know that, while I do like a good debate, I put great pride in being as fair and honest as possible, and I value truth and transparency before all. A number of users (and excommunicated users) have already experienced the Manichean, friend-or-foe approach of Twine's CEO, that's conditional upon paying lip service to the application. I have already denounced this type of PR approach in a post on this blog. Anyway, the point is that Nova mentioned he wouldn't take any interview with me anymore (he probably missed that I'm not a journalist), and so I have not contacted Twine for this post. I'm sure they'll respond one way or another, and hopefully we'll get some clarity as to whether this traffic issue is real. 

Let's be clear, I have also long complimented Twine and described it in my talks and posts as the flagship application of the semantic web - and I still do. This flagship status is why I am so interested in it. I believe we are only at the dawn of the metadata age, that our capability to create and manage metadata is still primitive and, as a corollary, that autotagging of the kind Twine achieves is much needed. I long for some of the functionalities announced by Twine for the end of this year, i.e. the ability to manipulate tags through ontologies and create complex queries as a user.

My theory was, and still is, that twine has a usability deficit. So many users going to Twine report 'not knowing what to do' or 'what it's for'. It lacks a sense of direction. Another example is the Daily Digest that a user receives from the application. I relied on this digest when I had but a few subscriptions to twines. Now that I have well over a dozen, the digest gives me an indigestion. Twine offers the option to select which Twines you want to see in your digest, and that's a step in the right direction, but not a solution to information overload: it's obvious what's needed is item-level selection and prioritization, not removing a full twine all together. As far as I can see, many other users have complained about this, but I was told that's a small issue and that's not the focus of the efforts. Addressing this, Twine would have the potential to become the filter of choice for information on the web, but I guess that doesn't strike them as something important to do. 

That wouldn't be a problem if at least Twine wasn't reacting defensively about those suggestions. But instead, interacting with them feels like grinding metal forks with your bare teeth. As such a rich tagging application, it's disappointing that Twine doesn't go beyond "good" or "bad" when labelling public feedback. Now it might be paying the price of its complacency towards constructive critics. And if the graphs above are confirmed, they might want to pick up a copy of Crossing the Chasm on their way home, too.

Related articles by Zemanta

• Twine is growing up (and up, and up) (joshdilworth.com)

Excellent slides on building buzz into your website

Designing For Social Traction

View more documents from Joshua Porter.

Linked Data, a Brand with Big Problems and no Brand Management

This post builds on the discussion that started on Twitter between Paul Miller and Ian Davis and quickly expanded to a few other folks. Paul subsequently wrote a blog post Does Linked Data need RDF?, and so did Ian Davis with The Linked Data Brand.

My turn:

I like Davis's suggested approach of looking as Linked Data as a brand.

It made me realize something: this brand hasn't been properly managed.

At this point in time, it's clear that beyond its inner tech circle, Linked Data is more often perceived as a technology that has wasted a lot more R&D funding than it should have, than as a solution to a problem.

If you are a brand manager, right there you have a huge perception problem to manage. In fact, at most companies, brand managers who let such an image problem spread out of control -like it has- without doing something about it would be terminated.

But there is no brand manager to sack for Linked Data.

Indeed, from what I've seen, there has been no deliberate effort from the W3C to manage that brand. That's where the problem lies. As Idehen illustrates so well with his comments to Ian's post ("marketing types will attempt to usurp and redefine the brand"), marketing is something many at the W3C have a 19th century view of, fed by Hollywood-propagated snake-oil cliches more than anything anchored in an actual education. In that regard, complaining that your brand will be re-framed is a little hypocritical when it wasn't managed and framed properly in the first place.

It may be time for the W3C to review its credo on marketing and do a little more of it. Time to learn it's not all coupons and spam. Marketers are also trained to align product development with real market needs (herein might lie the W3C resistance to it... want a reality check anyone?) and to -yes- manage brands. Even at nonprofits! I know, it sounds crazy, doesn't it?

So, as a "marketing type", let me try to make some constructive suggestions on managing the Linked Data brand, to tie this back to the comments left on Ian's post by Eric, William, and Paul, whose views I fully agree with.

First, and most importantly, pushing "Linked Data" as a monolithic block of technology conveys the perception one needs all the pieces to make it work. 1. That's not true. 2. That's enough to scare lots of potential users away, a bad adoption hurdle in brand management.

Instead, tell people you can include URIs in any piece of data. They may not get the full benefit of "Linked Data", but they'll get some of it, enough to start. In marketer's lingo, get them hooked. That doesn't make you a hooker. Doing that will get more people to try some pieces, see their limitations and find ways to address these, leading them to experiment with other parts of the stack and possibly alternatives to them. If RDF is as superior as the Linked Data apostles proclaim, it should win this battle, shouldn't it?

But Ian says he worries that solutions that only adopt part of the stack will not work well and create a perception that Linked Data is deficient, impacting the "seal of quality" he says Linked Data represents.

C'mon... 

For most, Linked Data rhymes with clumsy, academic R&D much more than with quality, as I wrote earlier.

If you really needed the whole stack for Linked Data to work, you would have a bigger problem on your hands: now you'd have to explain to users that to try Linked Data you need to understand and use RDF, URIs, SPARQL, OWL, and the other pieces of the stack. Good luck with driving adoption that way... Oh, but wait, that's what has been done so far, with the slow results we know of.

That's even why RDFa was created and is more successful (despite the confusing name). It's simplifying Linked Data into just one solution, easy to understand, easy to implement, easy to adopt.

Truth is, a lot of the advantages of Linked Data can be derived from using URIs or RDF independently, and lots more will come from mixing them with existing technologies to address real needs. Having them work together might be better, but it's by no means a must-do. That should have the Linked Data purists feel good about the whole thing. But no. Instead, they try to bundle up those technologies like pop and chips at a movie rental store (those snake-oil salesmen, some might say :)

Second suggestion about brand management. As we all know, it's very unclear what problem Linked Data actually addresses. Ok, I'm able to link data. Big deal. Most people after that ask "But what does it really do?". What does that mean for them? Well, in fact, there are plenty of benefits to being able to link data like that, but the people supposed to explain those can't do it in simple words. They suffer from the knowledge curse. They haven't been trained to communicate to non-rocket scientists or even mainstream programmers/webmasters, for that matter. They don't make things simple, and they don't make them sexy. That alone is a sin in this new world.

Case in point: calling it Linked Data makes it sounds like it's coming straight from the R&D lab. Bad brand name choice. Over time, you can address that, but you've got a hurdle to surmount right there.

So, should you re-brand or should you fix the existing brand? My guess is that a bit of both is happening right now.

On the one hand, we have companies making linked data theirs, putting their name on it, repackaging it. I think it's Peter Mika who said in our July semantic web gang podcast that what we called "Semantic Platforms" really were rebranded triple stores :)

It's not about making false claims and hyping it. It's about getting the word out about the real virtues of Linked Data.

For example: "With Virtual Knowledge, you will be able to quickly assemble your underlying data into custom interfaces that talk your language and meet your daily information needs. No more finding a needle in a haystack of news, no more calling the IT department to program interfaces or connect siloes, no more not doing it because it's too hard. You create your own dashboards in a few clicks and can see it all from there."

Or even "With RDFa, you easily embed information about your content. It helps both search engines and portals better understand and index what your data is about. This way, they can return results that are better structured, better presented, and thus more attractive to users."

Isn't that better than "With Linked Data, you interconnect the data through RDF, which creates qualified links. Those links can point to URIs, giving concepts unique central addresses, and add semantic granularity to your data. Add some SPARQL to run complex queries and OWL for the ability to create robust standard taxonomies, and you have your  recipe for success. Of course, you should really use all those together for the best results." Mmm, I can't wait to try that...

Now, as I said above, I think in spite of it all that the Linked Data brand itself is in the process of being fixed up. Even the toughest Linked Data fundamentalists have shown some signs of easing up on their purity claims, and understanding the importance of better marketing.  More importantly, we see more companies, like the New York Times, who want to "publish Linked Data". But what's most transformational is that use cases are getting clearer, for example with the adoption of RDFa by Yahoo and Google. Note, it's not pure W3C Linked Data, but hey, it's Linked Data nonetheless. Just like Coke Zero is still Coke.

There will always be people who argue that the original Coke is the only true Coke. Meanwhile, people drink the one that fits them best and still call it Coke.

Related articles by Zemanta

• Does Linked Data need RDF ? (cloudofdata.com)
• ReadWriteWeb Interview With Tim Berners-Lee, Part 1: Linked Data (readwriteweb.com)
• Google supports RDFa - SEOs work changes forever (ithoughts.de)

Attention Web 3.0 Start-Ups: Redirect 25% of your R&D Budget to Market Understanding and Development

In my observations on the Semantic Technology conference last month, I mentioned I would be blogging separately about the dire lack of investments by Web 3.0 start-ups in market understanding and marketing.

Before I start, let me signal some conflict of interest here: in the course of my job, I lead efforts on market understanding and positioning, brand awareness and lead generation. The idea is that developing my market is in the interest of start-ups with a stake in Web 3.0 and the semantic web. I get more business when they do :) But please do discount my words for this bias, and decide for yourself, when all is said and done, whether we have a solid case here or not.

My main claim with this post is that companies in the Web 3.0 and semantic web space are downright bad at filling unmet needs, packaging their offer effectively and cutting through the market noise to secure users and revenues. It even appears this incapacity to grow revenues has a direct impact on their ability to make money (ok, it's sunny outside, after five days of clouds there is no harm in having a little fun, is there? ;)

The large and growing number of players at SemTech told me that we are already in a crowded market. And yet: few real products, and very few sales. Most of the people there live on R&D funding from public institutions and overtrusting angels, not client receivables.

Why?

As a VC said during his talk, "this is a very early-stage market".

Allow me to translate.

That's code for "I don't get it, what are you guys selling here? Better fix that quick if you want me to come back next year"

This has been an early-stage market for over 8 years, which by my estimate is 1,302 web years. And it is going to remain so, as long as players prefer to hire one more Java "guru" as opposed to investing properly in market research, a business model and some marketing.

It was obvious, just going around, attending presentations and talking to people, that the business side of those companies is generally in an advanced state of starvation.

So many companies that presented couldn't enunciate any reason to use their products in an hour, let alone an elevator pitch. Too many had inexpressive and even customer-hostile names (ask me for examples by email...) and geeky logos that don't talk to their audience; too many, to present their product, sent technology experts apparently missing the "communication" function; too many didn't have a clue about their market, how to penetrate it, and no thought as to why, when and how often would anybody use their "solution". Please, let's call it a technology.

If Web 3.0 companies are eager to become mainstream, it didn't show in San Jose.

I brought up the problem of underinvestment in market understanding and marketing up at my talk as well as with a few start-up executives. The most common response was that resources at start-ups are limited. I know they are, but I don't buy it when I see the same companies already employing a dozen folks working on the product development side and hiring extra R&D FTEs. What's the ROI of that as opposed to taking one dedicated business person on board (and it doesn't even have to be full-time to start paying off)? I can recognize a case of feature and scope creep when I see one, and this is the supersized version of it. "Just one more feature!"

The reason for that are diverse, and have a lot to do with the academic and R&D origins of this stuff. I'll expand on this in a broader post on my other blog To Revenue at some point, and explain how I think even fundamental R&D is moving towards a more market-driven approach. Suffice to say, most if not all of the investment for the semantic web went into R&D and technology development to build the tools (standards and applications), and nothing, or very little, to test and refine our initial understanding of the market problems these technologies would solve, and evaluate the best way to solve them.

Luckily, the solution is simple: as a good rule of thumb, I’d see every start-up in the web 3.0 and semantic web space redirect at least 25% of its R&D expense to consumer research, brand awareness, partnerships and marketing. All good things that will make a start-up stronger if they don't kill it (better that than a slow, painful agony).

As I see it, the development of any market solution is an exercize in bridging the gap between two things: technological possibilities and market needs. Focus on one at the expense of the other, and you'll find yourself quickly in la-la land.

I can't keep count of the numbers of start-ups I come across that have gone too far along the product development curve and find themselves with a technology they don't know how to sell and to whom. Invariably, they haven't tested the waters prior to funding  development or continued to tune that development to meet real needs.

Why risk a reality check? It was just all too exciting at the time.

So curb your enthusiasm and get yourself a suit, with that free business budget you just found.

 

PS. Business folks need to explore ways to deliver more value, too. It takes a special kind to be effective at a tech venture, and the rapid changes in marketing also make it challenging to deliver business results if you come from a traditional marketing background. I talked about the difficulties in balancing the tech and biz needs and cultures here, and I will tackle the marketing revolution in a future post.

Video of the Semantic Web Gang at Semantic Tech 2009

Here is the video of the live podcast we recorded at the Semantic Tech conference, courtesy Semantic Universe and Paul Miller who arranged to have this recording.

The sound is a little lacking so plug your earphones to hear better. An MP3 will also be posted soon by Paul is available on the gang's page.

Disclaimer: I am not responsible for putting my own pic on the video screenshot!

Lastly, my company name was misspelled "Growthrate" in the introduction, which is funny. Maybe I'll use that if I launch some economic analysis subsidiary :)

PS. you can also find an MP3 for the panel I moderated on Business Models here

The Semantic Web Gang looks back at SemTech 2009 from Semantic Universe on Vimeo.

Will We Soon See a Rally on Web 3.0 Start-Ups? My Take on Semantic Tech

About 3 weeks ago, I left a warm Canadian sun to fly south to San Jose's chilly weather, and attend the Semantic Technology conference. Or, let's be fair, 1.5 days out of the five the event lasted. The rest, mostly, I followed through Twitter, which by all accounts tends to be a little like watching the Olympics on TV, as opposed to being in the stadium: you don't get the pop corn and hot dog breeze, but in the end you know more about the games than people who were there (and not on Twitter too... it gets quickly confusing.)

Of course, the big objective at "Sem Tech" is networking. Thanks to the nice folks of Semantic Universe (thank you Eric and Steve), I managed to make it, last minute, onto the exclusive list of VC Connect, a cocktail reception which gathered both fundraising CEOs and investors (of whom more attended than I thought there would be. I guess the free hors d'oeuvres did the trick). I stroke most great conversations there (apologies to those I met outside of VC Connect!)

So, my impressions of the event. I know those are late but I'm not competing with news organizations, just with analysts, so that's ok!

Overall, it was sort of "hot and cold": there were lots of very positive signs and, by all admissions, there is still a lot to do before Web 3.0 and the Semantic Web truly happen.

Key mental notes I made to myself:

• Fully endorsed: all the godfathers are now at the table. This event marked the first participation of Google, confirming the company's mood reversal after the tough words Eric Schmidt had for the field a while back. Anyway, what matters is that the 800-pound gorilla now sits next to Yahoo's SearchMonkey and King Kong Microsoft, putting their seal of approval on the effort, and all the credibility that comes with it.
   
• Market traction (which some call hype): attendance was up by 16%, which, in these macro-economic conditions, is no small feat. Lots more companies are asking what they should do to get ready for web 3.0 and the semantic web, and visiting the conference to understand. I don't know whether they got the answers they hoped for, but they were there and that's one first step!
   
• Real products in B2B: there were some real, commercialized semantic products, especially in the B2B space, such as those of Cambridge Semantics, and products in that space are starting to use Linked Data and integrate information into the Semantic Web.
   
• B2C disappointment : on the consumer side, of course there was Wolfram Alpha, which is certainly interesting but ultimately not earth-shattering. But at the exception of search engines, I found that little was new in B2C. SIRI was presented as a "game-changer", but with all due respect to Tom Gruber, I believe it is way too early for "Virtual Personal Assistants" to take off. Technology is not ready to support such a value proposition, all the hype will only help create expectations that can't be fulfilled, and because of this it is likely to fall flat in the market.
   
  I saw nothing that took advantage of Linked Data (URIs) and the stack (RDF etc), beside those few apps we already know about, i.e. Zemanta, Glue... Even then support is limited (yet to come for Glue, I believe). Clearly the consumer space is still quite a bit in R&D mode (yawn...), and what I came across felt somewhat unfocused and confusing too. Hopefully more will happen on that side in the coming months, with the technologies maturing.
   

Sem Tech also showed me there were still challenges, at 4 levels:

• Incomplete technology : linked data now seems mature enough to create commercial apps, but as I described in previous posts there is a huge risk of GIGO (garbage in garbage out), as text analysis is lagging behind. Also, the general focus on aggregation is diverting resources from creating smarter application from a user standpoint. Nothing I saw at SemTech made me feel much better about those issues, although I believe there are efforts taking place in text analysis that may yield good results going forward.
   
• Usability deficit: way too many apps are still trying too hard to have machines do all the work. A majority of CTOs have a hard time internalizing that machines are still stupid. They are. AI won’t be here for another few decades. In the meantime, companies should focus on higher usability.
  
  I was delighted to see this theme emerge strongly at SemTech, when the search engine panel advocated turning the search experience into a "conversation" with the user. By that they meant asking the user for clarification when there is ambiguity, for instance, or too many answers to be practical. Beside improving the user's experience, dialoguing with her and collecting the results intelligently would really enable the creation of rich semantic engines over time.
   
• Underinvestment in market understanding and marketing : this is worth its own post... Stay tune, I'll publish it in two days. Let's just say this is a huge issue in the space, and while I witnessed a certain progress at Sem Tech this year, we are faaaar from the levels required to make this R&D haven a business.
   
• Forget VCs, unless...: I was also hoping to see significant growth in the number of VCs and strategic investors at Sem Tech this year, and yet there seemed to be very few of them around. I could count maybe a dozen on the participant list, and met only about five of them. Interestingly, all said they would not invest in anything “semantic web”, but in solutions to problems, and they kept repeating that they saw the main problem with companies at Sem Tech as being a lack of focus on a consumer or company problem (see, I'm not the only one!)
   
  I am not exactly sure what to think of that, though, coming from VCs. On the one hand, I obviously agree with them about the prevalent lack of a market-centric approach in the semantic web space. Yet, on the other, there is a number of companies with great semantic technologies just waiting to be properly positioned and packaged by a clever business team. What do VCs do if not helping with that?
   
  After all, if you ask them, VCs don't invest in "Web 2.0" either. Yet in the meantime, they are still funding the next social network...
   
  As was pointed out to me by a Gold sponsor of Sem Tech, even the best of VCs, such as my friends (before this post, that is!) at Intel Capital, end up sending confused signals to entrepreneurs: case in point, while professing on the Sem Tech podium to be only investing rationally in companies benefiting Intel, Intel Capital put money in companies with very fuzzy problem-solving potential and a very, very diffuse contribution to Intel sales. Such as Bragster, that I can't call anything else than a stupid social network, very unlikely to return any money to its investors. If Bragster is helping sell chips, what isn't?!
   
  So, when VCs don't follow their own advice, how do you blame entrepreneurs for not doing it either?
   
  But there is "hope".
   
  Since VCs pretty much invented irrational exuberance and benefit from it (believe it or not, I am fond of the VC model - I just wish there was a little more self-discipline in the field), we could very well see a rally on the Semantic Web and Web 3.0 this year or the next, that would of course temporarily put aside question marks around "market problems" etc... VCs need focal points like that, and since cleantech investing seems to be taking a breather, the web 3.0 is the next best candidate.
   
  Having said that, I am not encouraging any company to forget about tackling a real market need that they can get paid for. In any case, the best path for many of the start-ups at semantic technology companies is not to bet on a VC rally on Web 3.0 and rather to assume there is little investment money in the field. In other words, forget fundraising, and focus on revenue and profitability from day 1. If you do that properly, VCs may soon come knocking at your door (assuming they recover from their financial crisis wounds), and you will be ready for any rally.
   

Such minor considerations apart, it looks like things are moving in the right direction, although still a little too slowly on the business side. At Sem Tech, I was a bit let down by the limited range of new, innovative solutions solving a widespread or even a narrow problem, but I think that the stars are slowly aligning and we may see an acceleration wave. Especially if rapid improvements are made in technologies to structure unstructured data, if an increased focus is placed on revenue model, usability, and market awareness, and if you follow all the advice in this post and refer it to 10 other people...

During the live Semantic Web Gang podcast we recorded at Sem Tech, which will be available here when it is published, I said that business people should come in droves to Sem Tech next year and take things over from tech folks. While this was deliberately provocative, I think there is a huge imbalance in the field at the moment and it incumbs upon CTOs and other tech CEOs to rebalance their efforts, open the gates and welcome more business types in. Opening up the semantic web is, in my mind, the key to success. Rendez-vous next year to see how well start-ups have fared.

Related articles by Zemanta

• SemTech2009 impressions (ivan-herman.name)
• Siri: A Powerful Virtual Assistant For The iPhone (techcrunch.com)
• Google and Future of the Semantic Web (mehmetalierturk.com)
• Google embraces exercise in Semantics (news.cnet.com)
• Understanding the New Web Era: Web 3.0, Linked Data, Semantic Web (readwriteweb.com)

Tech vs. Biz Type Divide: Tackling the Number One Product Failure Driver Headfirst

I am back after quite some time, my ability to blog being constrained by business requirements. I am indeed happy to report that Growthroute's activity is growing rapidly, with clients in the web 3.0 space in particular. I know, I know, I have not even published my thoughts about semtech yet. I was there for less than 2 days, but there were some interesting things I took away that I have not seen expressed anywhere yet, so I will blog briefly about that very soon.

In the meantime, I have just published the slides of a beta presentation I gave at Communitech, the innovation and entrepreneurship hub in the IT mecca that is Waterloo, Canada -- and I believe it should be of particular interest to ventures in the semantic web and web 3.0, since it tackles the gap between techies and business people, discusses how that affects commercial success, and explores potential solutions. The post and slides can be found here. I welcome all constructive feedback, with few limitations on what constructive means.

Excellent Video on Web 3.0 and the Semantic Web

The following video was pointed out to me recently and I find it to be a terrific overview of the current evolution of the web:

I got this from here. I don't know about the European Union claiming to spearhead the next web but, that detail apart, it is good material.