What happened to Web 3.0 — the semantic web?
, 2,467 words
...or "Why isn't the future happening"?
Despite over a decade of discussion, "Web 3.0" -- the next model for the Internet -- has not happened and is not about to happen any time soon. Why?
Concepts for the social and semantic web were created early
Broader access to the Internet came around 1993. This first incarnation of the web was largely read-only, with fairly simple mark-up, little dynamism and browser incompatibilities which propelled use of Java Applets and Flash. Notable uses of the technology were directories, portals and search engines.
By 1999 the term “social web” (or Web 2.0) was coined. Conceptually, represented a fairly simple enhancement over the first iteration of the web. The idea was that sites would be read-write: users could contribute to pages and the interface and user-experience delivering this would be rich. This technology could be used to deliver personalised content, with users interacting with each other and with content creators.
In parallel with the social web, the groundwork was being laid in 1999 for the “semantic web” (or Web 3.0). Every device would be connected to the Internet, and devices and sites would expose their data in a structured way to be “mashed up” and reused elsewhere. This would form an Internet of Things, where information could be simply, reliably and automatically extracted and applied to different contexts. Mobile devices would be able to retrieve and reuse information such as pricing or review scores from the web, fridges could automatically reorder groceries, and cars could drive themselves.
Whereas the first web served information for use by people, this third web would be built on contextualised data and metadata which machines understand, trade and build on. Machines interpreting data in this way underpins the popular Silicon Valley vision of the singularity, and of future where autonomous software agents can understand and make decisions on behalf of their owners.
The dotcom crash and the birth of the social web
Fewer than five years after being thought up, the social web was alive and healthy, enabled in the most part by the dotcom crash. Throughout the boom, development had been constrained by the performance, flexibility and cost of proprietary technology. Monolithic systems from Oracle, Sun, Vignette and Microsoft (amongst others) restricted functionality, prevented innovation and drained budgets. Whilst few platform companies went bust directly, the crash hammered their market share and drove users to Open Source technology.
Open technology and standards let developers build applications for tens of thousands of dollars, in contrast to the millions required before. This extra attention attracted expert developers and companies such as Google who made significant commitments sponsoring development, and opening up their own technology.
At the same time, Microsoft's nascent focus on internet-ready desktops and their part in the browser wars were shaking up Internet Explorer, enabling vital techniques such as Ajax. Non-web internet protocols dwindled in popularity, with technologies such as ICQ and FTP being replaced by web-based social networks and content delivery networks. Java Applets rapidly died, and Flash began a slow and insecure death, lingering on until the iPhone finally killed it.
As Web 2.0 matured it was reinforced by fundamental changes to what could be done with simple pages. HTML5 enabled delivery of video through the browser, and CSS3 brought responsive design, enabling desktop sites to render cleanly to mobile devices. Newer internet heroes Igor Sysoev and Roberto De Loris built on Linux, creating Nginx and uWSGI, which let developers create rich, event based software-as-a-service ("SaaS") applications for only a few thousand dollars resource cost. Today's web has come so far that the new trio of social, mobile and local applications (SoMoLo) is sometimes referred to as Web 2.5.
So why no Web 3.0?
So, the inevitable question. If both 2.0 and 3.0 were designed at the same time, why has only one of them been implemented?
In part this is because the transition to 2.0 was necessary and inevitable. The first internet giants such as Sun, Oracle, Cisco and the telcos were able to punitively extract money from pioneers building sites. 2.0 turned this on its head, decimating value-extracting platform providers and moving power to the companies running the sites. Some companies – like Yahoo! – floundered, but Amazon, Google and others were able to embrace Web 2.0.
The new giants -- Google, Facebook, Amazon -- offered valuable, pervasive services to their users for free, or at almost no cost, and were not beholden to closed platform providers. Users took to blogging, writing wikis, podcasting and commenting on sites which were more dynamic than ever, and site owners took advantage of revenues from publishing, transaction and advertising systems.
Web 3.0 levels sites which exploit poor distribution or accessibility of data
Whilst the second iteration of the web offered benefits for user and site owner alike, the benefits of 3.0 are asymmetric. Using the web as an API to create sites and services with easy access to data is great for consumers, new market entrants, and for the few giants that sit on top of it all. For the vast majority of transactional sites, however, these benefits would come at a cost: sharing their data removes a barrier to competition.
This new web would bring with it the ability to uniquely identify content, assess data accuracy, and uniformly compare like for like pricing. These three are enough to level the margins on most gambling, ticketing, and commodity ecommerce. Sites generating revenue from or contributing to disinformation or inefficiency of information distribution would be wiped out or bypassed.
As an example, the musician Watsky was playing in London this May. The venue's site sold tickets directly for £11.25 and had plenty available. The top Google results for the gig were for TicketMaster/LiveNation and ViaGoGo. The former requires a 4-page journey which involves typing "sustorsc roonov" into a Captcha box before one can learn no tickets are available. The latter offers some for £56.82 each, with a further 15% customer service fee and £10 for postage. That is a 670% markup.
These businesses, which trade around fear, uncertainty, doubt (FUD) and misdirection as pertains to information, would be wiped out by Web 3.0. Use of FUD as a sales strategy brought a lot of the value for Web 1.0 platform companies; these days it's Web 2.0 vertical companies which use it to extract outsize margins.
The third iteration of the web could enable one's phone to quickly and simply tell whether tickets were available and buy them without having to visit a multitude of sites, pay a premium to an aggregator, or wrestle a Captcha form. A software agent might, on one’s behalf, identify free time in the calendar and automatically find, book, and arrange relevant events.
Without 3.0, many companies are striking out alone in narrow verticals, trying to build technology which works as if 3.0 had been realised. As examples, Skuuudle are a small company trying to make price comparison easy, Hipmunk are rather more advanced and are doing flight comparison, and Google are famously pulling data together for their fleet of self-driving cars. Technology like this will only pervasively be available out of narrow verticals with Web 3.0.
Are there other iterations of the web which might be implemented first?
Seth Godin writes off the third web as unrealisable, and suggest a Web 4.0 where the semantic agents of 3.0 are fudged to work on the imperfect data available now. However, at best that's Web 2.5, like Om Malik’s "alive web", the “card or canvas web”, or the “web squared”.
Others have tried to redefine the current web as 3.0, and even the odd over-excited venture capitalist has tried to redefine 3.0 as "mobile" or something equally daft, before getting shot down in the visitor comments – a fittingly 2.0 result.
It is likely the big three’s platform offer will become more appealing for use by smaller players, but also more likely they will be consumed by it. The semantic web will arrive, though perhaps it is still five years away.
In the meantime, I'm long on vertical ecommerce and have had the pleasure of investing in and managing a number of businesses in the space. I hold and still make a number of investments there. Let's look at some of the practicalities.
What would Web 3.0 mean for established ecommerce businesses?
The levelling of information brought by the semantic web will power both new entrants and Internet giants to consume and redistribute information from vertical ecommerce sites. It is no coincidence that the platform giants have built marketplace, checkout/payment and product modelling systems. In particular, Amazon and Google will process payments, host servers, enable product resale through their marketplaces, and even -- in Amazon's case -- manage fulfilment for smaller vendors. Google and Facebook have even built OAuth login systems which can be used by site owners to let their users log in.
At the moment, these tools are largely opt-in. Ecommerce sites choose to list their products and pricing by supplying a special feed of data. With a semantic site, the ecommerce players won't need to maintain this feed: the aggregators, price comparison sites and platform giants will just be able to take it. Giving up this initial direct contact with potential customers by using a marketplace is the beginning of the end for larger ecommerce businesses. Just as ruthlessly as the supermarkets pushed their way up the value chain to destroy wholesalers and then squeeze producers in the UK in the 80s, and as vertical ecommerce sites squeezed wholesalers in the earlier 00s, the platform giants will rapidly marginalise ecommerce companies once they start exposing rich data. How?
The squeeze starts
Firstly, marketplace providers gain access to a set of particularly valuable information: they learn all sorts of things about which products sell well, at what price, to whom and when. As an example, Amazon were effectively given free data on the reach and margin two verticals and were able to swoop in to buy Zappos and undercut diapers.com until they reached a deal to buy the business. The change here is that the platform these ecommerce site operators build on is no longer the cables and servers version of the Internet of old, it's increasingly a serviced-based platform. This platform is cheaper, faster to scale, works better... and will compete with the same people who are building their sites on it. All of the big three, Amazon, Google and Facebook have a compelling offer for etailers: providing traffic and conversion opportunities in exchange for commercial, product and stock data.
The second and most significant challenge is the lack of control that site operators have over marketplaces. The marketplace sets the rules. Companies selling apps for iPhones complain about the difficulty of selling without being featured, and about Apple's 30%, but there's little they can do about it. The ecommerce marketplaces play out in a similar way. Companies using the marketplace must compete by different rules. Vendors selling commodity products will be shown to potential customers based on how well they integrate with the marketplace, how cheaply they price, and the sort of reviews they get. Existing differentiation on pure service, brand, delivery, descriptive data or imagery become some of many things to compete on. Ultimately, marketplace vendors will get penalised for not using platform-provided payment or fulfilment options, much as eBay do with PayPal.
How might vertical ecommerce businesses avoid this fate?
Unprepared retailers will do what suits consumers least, and not expose their data in semantic or easily parseable formats. Only a relatively small number of companies are willing to share their data: content companies such as Yelp or Zagat (bought by Google) who want their scores embedded in Google's search results using microformats, or newer market entrants trying to dislodge larger ecommerce businesses.
These businesses can head off the risk of being side-lined in a marketplace by giving users a reason to visit their site other than transacting. Transaction online is rapidly becoming a commoditised service with the likes of PayPal, Google Checkout, Amazon Payments, Stripe, and Square. The days of different players having their own or wanting to have their own transaction systems are limited. Instead, investment in systems which are harder for content players to replicate is worthwhile: retailers creating differentiating non-retail value! These fall into five categories outside of the usual sorts of things which ecommerce companies will be doing, such as providing great service and engaging closely with their customers.
Five techniques for building and maintaining a strong ecommerce business
These mechanisms help ecommerce companies stay high in the value chain, so that they can avoid being reduced to price/integration competition on a platform site’s marketplace.
- Invest in strong, unique product description, preview and imaging. Making the site’s inventory richer and sexier than it is on similar sites can be critical for keeping traffic and boosting showrooming. It is still relatively easy for smaller sites to do a better job of showing products than Amazon does. Relying on manufacturer or brand-supplied SKUs and descriptions is harmful in that it makes price comparison easier, and any generic content translations can be easily harvested by competitors.
- Generate stickiness on-site with buying, planning and complementary tools. Similarity, loyalty systems and elements of community give users a reason to remain on site or to return and potentially be converted later. Amazon has done well with Prime and nimbler, smaller companies can do a lot.
- Source own-brand & product personalisation options. With own-brand product retailers have the opportunity to create exclusive products and deals. Often similar to conventional branded products and originating from the same factories, they can provide tie-in and help the retailer retain control. Similarity, online product customisation systems are great for margin and cater for interest in makers and custom goods. Amazon have been quietly building a large inventory of products under their own brands, and even repackaging branded goods.
- Entrench with added value and alternate revenue streams. By offering price aggregation, comparison or marketplaces (both new for other retailers & second-hand for consumers), ecommerce players can to position themselves as top of their vertical. After Play.com’s strength in entertainment media was destroyed by a tax change, the owners managed to sell the business on the strength of its marketplace, which held up well in relative terms.
- Offer brand extension options to suppliers. Just as supermarkets sell aisle placement and banners, sites such as KitBag and CNET are masters of offering co-branding, endorsement, and customisation options to secure brand support in cash, margin, stock and supply exclusivity. Amazon are already adopting or countering many of these tactics head-on: they’re using partner sales data to expand their range of own/alt-brand “basics”, and will offer non-niche brands a better deal direct through their own platform, packaging, logistics and promotion.