Photo credit: Peter Galbraith
Picture this:
- A browser that instantly shows you the content you'll find most interesting
- Search engines that return fewer, better results - every time
- A marketplace that always tells you the best products and services, and lets you advertise anything you like
- A world with no more spam, phishing or online scams
- Being able to access your contacts' current details all the time, without effort
- and being able to find the right contact wherever you are
I'd like to share my vision for how all this will work, which is low-tech, doesn't rely on any new technology, and is nearer than you think. And I predict that Yahoo! will probably be the company to make this happen, in the next 2 years.
An evolutionary step
The web has given us some great applications that let us do some amazing things: communicate, publish, share, and shop with greater ease and speed than ever before. People, content, and data are more fluid and connected than ever before.
However, the world-wide web has now reached the limits of the platform.
These great applications that we have are still fragmented and inconsistent when they could be interconnected and co-operative.
The new web will be a platform that simply joins up all the functions and data with a new network of connections below the application level, enabling a new generation of applications that will make life easier, more productive, safer. and more fun.
Searching today's fragmented web
Think how often you use an internet-based application to find something: content, recommendations, shopping, services, contact details, research, comparison...
Remember: "The answer is out there!"
The right answer is nearly always out there - somewhere.
- That person's contact details do exist online
- There is a piece of software that does just what you need
- There is a company that sells the item you need at the right price
- Someone is trying to sell the sofa that's perfect for your room
Most web applications are trying to match information to where that information is needed
The major problem is intrinsic to the Internet's architecture: there is no center. The lack of center is the reason why the web has become so useful so quickly - it has enabled free, organic growth.
But it also means that any time we try to compare information, we have to try to collect information in one place - such as on a single database. That's difficult on the web, although there are partial successes, see below:
Examples of the fragmented web
- Ebay has a huge database of people selling stuff, which it matches to people looking for stuff
- Google has a massive index of web page content, which it matches to search phrases
- Friends Reunited has a database of where people went to school
- Car Harbor is a solution (currently in development) that will match people looking for car parking with people who have free space to rent
- Loads of dating sites have databases of people wanting to find people who want to find people
- LinkedIn has a database of businesspeople's information, and links between them
I could go on. What these applications have in common is that:
- They all try to match information with needs-for-information
- They all have their own databases that contain a small proportion of the information out there
- And their scope is limited to only the data they hold themselves
In other words, they all have part of the picture.
This means they can only try to find the best answer to what you need with the information avaialable.
It doesn't matter how much bigger or smarter these systems get, they're limited by the fragmented web version 1.0. Google or Ask.com will never be able to know what you really want when you search for "home run". This is because the current web is still locked into reductionism. Because all these applications are just part of a disconnected world of data, they're forced to reduce everything to their basic component parts.
Case study
You want to find someone to come in and help clean your house.
Now, there's someone living half a block away who would be perfect for the job.
But...
- They won't have a web site advertising their services
- They won't be in the Yellow Pages or on other listings guides
- They promote their services only via word of mouth
- They have done similar work for a neighbor of yours whom you don't know
So there is a right answer out there, but how do you find that person?
If you use a search engine to search for "Cleaner Chesterfield", what do you get? Absolutely nothing of any use! Google returns about 248,000 results, which are useless to me!
I don't want 248,000 results! I want one good one!
The search engine has taken the information I provided, and it has done its best. I don't blame Google - its algorithms are very powerful. It's just that power isn't enough to solve this problem.
What's needed is a new way of looking at the problem - a new system that appreciates the context of my query, not just the words I use.
One-dimensional search
Today's search engines take the whole rich universe of information, which contains my problem (query), and a plethora of possible answers, all floating in a multi-dimensional space of context.
What they do is squeeze that universe of possibilities into a one-dimensional list of 248,000 results.
They treat all information as equal, which is missing the whole trick, because all information is not equal - to me, here, now!
Visions of a more connected world
Photo credit: Peter Galbraith
What we need is to connect everything in such a way that, from a single entry point, we really can find the best answer to our information need.
I'm a software designer, and I think best in terms of how solutions feel to use. So let me describe the future web through user experience stories.
I'll describe to you my vision through the following examples:
- a future killer homepage
- a future killer marketplace
In two upcoming articles I will complement my vision by describing to you other components of the future panorama I see, including:
a future killer search
a future killer contacts manager
future killer security features
A future killer homepage
I open my web browser. Instead of showing me the latest news, or summaries of my RSS feeds, or some combination of these things with a bit of weather and stock data thrown in, my browser simply shows me: what I'm most likely to be interested in.
How about that? No messing around, no settings and options, just what I'm mostly likely to be interested in.
How does it know?
Well, the answer is out there, isn't it? There are things on the web today, which may be new, that I'd be really interested in if I saw them. All the system has to do is work out what those things are, from what it knows of me.
Now, there are already all kinds of data about me on the network. The problem is that, because of the disconnected system architecture, these bits of data don't know each other. In the connected web, we simply join all this stuff up in a way that magnifies the usefulness of each bit.
Let's think what kinds of thing we might know:
- My profile - age, location, gender, socio-economic information
- My bookmarks - in my browser bookmarks, history or sites like del.icio.us which indicate what information I'm interested in, what sites and companies I prefer
- My connections - on MSN, Yahoo! Messenger, Skype, email, LinkedIn and Yahoo Groups etc. - which indicates who my friends and business contacts are
- My words - on my web site or blog
Let's start with just these. Say we could join them all up into a single system where all the data could be viewed and used together. What might we learn about me?
The power of connections
Aside from the raw data about me directly, we can learn quite a bit about me from the sites, people and groups I'm affiliated with. If you know my favourite sites and favourite people, you can start to build a picture of what my favourite stuff is.
(I'm not talking about any fancy fuzzy logic or AI here, this is pure number-crunching at work, and I'm not talking about any futuristic technology.)
Think of it like a heat map of the web.
We're only interested in the things I'm directly associated with, and the things my friends and contacts are directly interested in.
Once we have a heat map of the areas that most interest me, we can look at key hotspots on the web to find the hottest content according to my profile for today.
The things I've most recently expressed an interest in (such as new trends, new sites, new contacts) may be hotter, and these things might "cool off" over time.
Similarly, things that I've repeatedly connected to (through regular visiting), can be interpreted as more important.
The killer homepage will easily be able to predict the things that I'm most likely to find interesting: to do with interaction design, web design, the environment, politics etc. just by looking at a heat map of where my interests lie, and what's hottest in those areas today.
We're starting to see new things that become possible just by joining up our data.
A tool for collecting connecting information
Check out the toolbar below (which I've mocked up). It already has most of the tools we need to power the connected web.
- The "Rate" tool and Bookmarking let me record pages I like (although the browser can also simply record which pages I've been interested in by what I've looked at, for how long, and how much I scrolled)
- The Buddies tools will show me my list of contacts, tell me who's online, and let me review the strength of links I have with them and my permission settings.
- The green flag tells me that the community thinks the site I'm on is safe
- "Auto-login" is another usability and security feature that saves me time and prevents phishing.
Future killer search
Today's search engines are not great at recommending the best results to a search query because they are limited in scope and reductionist in approach. They treat all information the same, however there's an easy way to cut down the right results by half.
The easiest and best thing we can do to make searching for products and services far more powerful is simply to separate things that are offerings for products and services from everything else.
Just take all the pages with people offering e.g. fishing tackle, and deal with them on a separate Marketplace system (see later), and are left with everything that's simply talking fishing tackle.
I'd guess that about half the searches done on the web are ultimately looking for solutions (eventually in the form of products and services). In other words, we're looking for something that solves a need we have. The rest of the searches are for raw information.
The problem with most of today's search engines is that they treat everything as raw information.
The reason that eBay is so great, and so successful, is that it applies this simple filter. Ebay lets you search directly for things that you know people are offering. Comparison sites like Froogle apply a similar simple cut to e-commerce offerings.
By removing the whole supply/demand dimension into a separate Marketplace, we improve the quality of searches for product by maybe 1000%. Search engines can exclude all information relating to products and services, and can concentrate on doing what they already do best - matching search terms to raw information.
However, once we start joining up our world, the tasks of figuring out what content is most relevant to a particular search become so much easier!
Remember what we said about predicting what content will most interest me for my killer homepage?
Search uses the same technology.
- What have I looked at recently?
- What pages have I spent most time on?
- What content have I been reading?
- And what have my friends, contacts, and colleagues been looking at (people who are likely to be in similar businesses to mine)?
This kind of search isn't even that smart, yet it will be able to give me a shorter, better list of results.
Note: This approach is already being called "Social Search".
The key difference that the connected web concept brings is joining together all the entry points, so that it doesn't matter what means you use to access content (RSS subscription, del.icio.us, or your browser bookmarks), all the data can be accessed in your personal networked heat map when you do a search.
More to follow. Stay tuned.
About the author:
Ben Hunt, aka the "Web Doctor", is a UK-based interaction design consultant who has been designing software, web sites, and web applications since 1996. Ben runs a small UK-based consulting business called Scratchmedia, and publishes 'Web design from Scratch' to share his passion for designing excellent user interfaces with a worldwide audience. His key mission is to help make the web a better place by learning and sharing the practice of good design.
Ben Hunt -
Reference: Web Design from Scratch [ Read more ]