Content-type: text/html ~ Stephen's Web ~ What Not To Build

Stephen Downes

Knowledge, Learning, Community

Jan 04, 2009

Originally posted on Half an Hour, January 4, 2009.

I get to play a government scientist on the internet. As a result, a large part of my work involves being exposed to new and interesting technologies, whether they are the latest military simulators, academic papers delivered at scientific conferences, or product proposals being promoted by aspiring developers.

My sort of environmental scan is a bit different from what you'll get from consultants and venture capitalists. Don't ask me what companies are developing what products, how industry stocks are performing, or where all the 'smart money' is going. I don't know and I don't care.

What I can tell you, though, is what technologies are working, what technologies are flopping, and what technologies are fads. It's practical, down-to-earth advice. For example, if you are a technology developer, you already know that you should not try to build a new operating system, a new word processor, an online store or an auction site, for example. These have ben built and have established a mainstream presence. You would need thousands of engineers and billions of dollars to compete with them.

The rest of my advice is like that, only more nuanced. It's obvious to everyone that they should stay out of the operating system market, yet much less obvious that they should avoid building a new content management system. It's less obvious, because these things are harder to see, but it is none the less certain.

So, here is my advice on what not to build. Actually, it's a bit more than that: it's a list of what not to build, a list of some things that people are working on now, some fads to avoid, and some indication of what's out there for the taking, if you can get your act together in a hurry. And what lies beyond that? The domain of real innovation and progress.

What Not To Build

Don't build a destination website

People are still building destination websites. They expect to build *the* location to find such and such. It's somewhat surprising to see in 2009, given that every company, every school, every library, every museum, and every other organization, product, service and even many pets have websites. Even if you have an original website idea, your site, unless it is *very* special (like, say, the Dead Sea Scrolls) will quickly be swamped by the noise and verbiage that is the web, your only traffic search engines and spammers. Even if you have original content and original ideas, don't just build a website.

Don't build a CMS

Unless you have an established market of community and content web sites, you have no business building a content management system (or, for that matter, a learning management system). There is a wide range of choices for people out there, everything from Drupal to Blackboard to SiteScape. And people looking for hosted content can use Blogger, WordPress or LiveJournal. And even more to the point, basically every large scale operation that is going to want a content management system aready has one. You will be facing tremendous competition as every new and existing client will be choosing from a range of well-funded commercial and open source products.

Don't build a platform-specific app

2009 is likely to be a year in which everyone is building Facebook apps, Flickr apps, Twitter apps, iPhone apps and Second Life apps. But this is a market you want to avoid. For one thing, it is already saturated. Indeed, any time something becomes popular these days, it is designated as 'a platform' (in homage to web 2.0) and a horde of app-builders descend upon it. The platform remains popular for a while, but as it declines (as it inevitably does) it takes the entire set of platform-specific applications with it. And you risk, a any moment, the platform proprietor building competition for your app and putting you out of business (this applies to Android too, in case you're wondering).

Don't build a Java application

This is a bit of a special case of the preceding recommendation. Java is the original 'platform' application. What that meant was that you had to have Java installed (and, as time went by, the *right* version of Java installed). Java has been around for ages now, and yet most computer users could count on one hand (or, in many cases, zero hands) the number of Java aplications they use. The situation is a bit better now that Java is built into some operating systems. But java's day has come and gone - everybody is into the platform-building game now, and most have learned from Java's mistakes.

Don't build a framework

This is one of those bits of advice tat would not apply if you could actually do it - that is to say, if you *could* build a framework, then it may be worth doing, but for the most part, you probably can't. It's advice the Perl Parrot and Radaku projects should probably have heeded ages ago, advice the Ruby on Rails people should keep in mind today. People involved in those perpetually running framework projects are tossing good money after bad. Basically, if you are working in any well-known computer system - Microsoft, Java, Javascript, Python, whatever - a number of frameworks already exist. Javascript, for example, supports a number of frameworks for doing web 2.0 stuff - JQuery, etc. Now - you may say, the framework doesn't do everything we want. Maybe not. But that's not the lesson. The lesson is, if a framework already exists in your domian, your domain has been commoditized. Get out, get out now.

Don't build an educational game

This bit of advice is pretty specific and probably does not apply to most people (since most people would not dream of doing this in the first place). But the question to ask yourself is, what is a game doing for you that a straight-forward presentation of the information is not? If it is specifically an *educational* game, the answer is, "nothing." You're not getting new users, you're not presenting material in any way that's easier to understand, you're not adding to motivation. You're simply disguising the old 'teach and test' methodology as a game. Nobody will be fooled well, except maybe purchasers of fad educational products.

Don't build a new standard

People are still proposing to develop, or work on, new standards, be they metadata languages, vocabularies, application profiles, and the like. back in the days when no standards existed, this may have been a good idea. But today, the standards landscape is full. There are standards for every domain under the sun. Things that probably should not have standards - like carrier pigeon messages - have standards. What's worse, few of these standards projects made any effort to work with or cooperate with existing standards. So the standards landscape is a mish-mash of convoluted over-engineered and competing standards. Unless you absolutely have to, don't add to this landscape. Work with what's there and extend it (even if the rules say you can't).

Don't build a new social network

First we had several dozen social networking sites, like Friendster and Orkut and MySpace and Facebook. These became platforms (see above) and then we had social network multiplier sites, like Ning. And now (so-called) social network websites are multiplying like, well, websites. These social network sites are nothing more than reworked mailing list sites (like Yahoo Groups and Google Groups) and content management sites. And the blog-based social networking sites, like MyBlogLog, have already been commoditized. The irony is, as he number of these social network sites increases, their usefulness decreases. How many people are now refusing invitations from new social networks? Right - that would be everybody.

Don't build a wiki

This is a special case of the social network site. A wiki requires a community of people to work together to provide a common base of content or services. In order for a wiki to work, the contributors have to massively outnumber the spammers and the griefers. This works well if (a) the site is sufficiently massive, like Wikipedia, or (b) the site is sufficiently obscure. The Wikipedia project could be duplicated a few times before the pool of potential contributors is sufficiently diluted. That time has ling since passed. Your wiki will be either (a) obscure, or (b) filled with spam.

Don't build a travel site

This is another special case. What it refers to is not the travel site specifically - though this market is saturated with the likes of Expedia - but the web services sites generally. The 'travel site' was always the paradigm example used to promote web services. But, just as it would be foolish to try to build another travel site, so also it would be foolish to try to build most web services applications. The point is, when you choreograph multiple applications, the market fills up very quickly. One travel site, for example, basically has a lock on hotels, airlines and car rentals. Web services sites are category killers, and most categories have already been killed.

What is there? Stuff everybody is working on

These are not things I would say you should avoid outright. The market is not saturated, there is room for innovation, and new products will be appearing over the next few months. But beware - a lot of people are already working on these things. If you have to start from scratch, you will have a lot of difficulty catching up. Your best bet right now is a niche play somewhere at the margins.

Alternative interfaces

Nintendo scored a huge hit last year with the alternative Wii interface and their success is drawing a lot of attention. People are now looking at all sorts of ways to control a computer game or computer interface. Webcam interfaces appeared a few years ago. Motion-sensitive and orientation interfaces are featured on things like the iPhone. I've seen gesture-based interfaces (with and without data gloves). I played a game with a heart-beat monitor last summer. And I've even seen a game based on a brain-wave detector.

Portability / cloud / smart cards

Cloud computing has attracted so much attention recently that it's a candidate for fad status. But behind the fad is a set of concepts that have legs - the idea of computational portability. By this I don't mean mobile devices (though obviously they play a role) but rather computers that can plug into other computers to allow you to move your data, software, authentication, and whatever else you want. We have smart cards in our credit cards now, but why can we have our web browser, email application, and social network in our smart cards? The answer is: we will.

Calendaring / coordination / events

There is a range of applications we might call Kantian applications - they depend on time and place. Historically, Apple and Microsoft have kep calendaring to their proprietary little selves, but this logjam is breaking, and calendar-based applications are becoming available in our personal lives as well as our business lives. Which is good, because everything from social events to concerts to television listing to anniversaries depend on time. Finding new uses for time - that's an opportunity that will not go away any time soon.

Location-specific applications

The second group of Kantian applications are those that are taking advantage of publicly available GPS to create location-based services. These should become widely mainstream as mini-GPS systems are built into cameras, phones, PDAs, laptops, cars, belt buckles, keychains, and more. I personally could have used a GPS based keychain locator this month - I don't even know what city my keys are in. And keeping track of children, vehicles and pets will pass from quirk to mainstream over the next few years.

Intelligent apps / recommenders

We want in our everyday lives what is already available in some aspects of our professional lives - the ability to pick the best product or service in a given environment. Expedia, for example, allows me to pick hotels quite efficiently, and while it can be fooled by unscrupulous proprietors, the service is getting better over time. No such system exists - reliably - for consumer electronics, for rental accommodations, for cars, for food. Imagine, for example, a system that created my grocery shopping list for me, so I simply didn't need to figure out what I needed and wanted. Or that reliably recommended (and delivered, for free) books and music. Moreover, there is not only room for an extended range of recommenders, but there is also scope for increasingly reliable recommenders.

Connected applications (walls, desks, fridges, toasters)

OK, maybe not toasters, unless you really value weather maps on your morning toast. But with ambient wireless in an increasing number of homes it has become feasible to connect appliances to the internet. This creates a whole range of possible products - paper-thin displays that hang on walls, desks with smart, interactive surfaces, fridges that keep track of your food, automatic light switches that switch off when the room is empty, health monitors, and more. And it's not just that these applications are connected to the internet, it's that these applications can access your data, remember choices you've made, and interpreted and project your needs. What needs? That is where the room is for innovation.

Sensor networks and sensor data processing

A lot of work is being done in the field of sensor networks these days. There is a number of obvious emergency-related applications: fire sensing, flood sensing, intruder detection. Weather reporting should evolve in short order from a small number of central weather sites to a dense grid of home and business weather stations - and these, in turn, by making their data public will allow businesses to better manage staff, stock supplies and anticipate markets. Sensors already manage the flow of traffic in cities, and will increasingly manage the flow of goods and people. Room for innovation here includes coming up with new things to measure and developing algorithms that analyze and understand large grids of related data.

Summarizing, data extraction, decision support / workflow support

Business intelligence services already monitor and analyze web and internet traffic in support of corporate and military intelligence. But there is room for personal intelligence services - wouldn't it be nice to know about that patch for your Zune, for example, before it suddenly freezes? And there is a need for people to be able to make sense of an increasingly diverse information space - especially as the traditional media can no longer be trusted (if it ever could) to describe events fairly and faithfully, or to report on obscure or unpopular disciplines.

Predictive data visualization

Data visualization will become even more useful when it becomes predictive. We already have a sense of this: we use predictive visualization every day in order to understand what the weather will be like (and despite widespread criticism our weather predictions are surprisingly accurate). Being able to predict crowds, shopping trends, stock prices, fads and fashions, and more, will become an increasingly lucrative industry. Imagine how Air Canada could respond if it had a reliable way to visualize the mess that would develop as holiday travel merged with a series of blizzards.


Green computing

Green appliances have been identified for a number of years with an "energy star" designation, a system that worked mostly because there was little business advantage to proclaiming oneself green. That has all changed. Consequently, titles and labels will be of little value as being "green" becomes a marketing ploy rather than an indicator of energy conservation.


The key sign that the iPhone is a fad is the fact that most of the attention being paid to it has to do with applications and games, not telephony. In addition, the market for iPhones saturated itself within a few months of its initial release: pretty much everyone who wants an iPhone has one. Finally, other vendors - and in particular, Research In Motion, which has survived American patent protectionism - are matching (and sometimes exceeding) Apple point for point with product and service.

Cloud Computing

The interesting thing about cloud computing is that almost nobody in the public knows what it is. This makes it ripe for fad status. But to survive, cloud computing will have to actually be used - and people who don't know what it is won't be using it.


Online instruction system

It has always been the holy grail of the e-learning industry: a totally automated system that manages instruction for you. There will be no end to the number of people who say teachers are indispensable, but if the social function of teachers can be replaced by community, and the informational function by software, then a stand-alone online instruction system is possible. And that's what we're seeing people try to build, step by step, with learning objects, competences, and the rest.

Distributed systems

The idea here is to have a thing - a concept, an idea - that rests on, and floats above, a non-specific computing environment. This was the thinking behind the connectivism course we ran last fall. The idea is that the 'course', via its constituent teachers and students, simply grasps whatever computing environment is convenient and available, creating communications channels between those environments, and hence establishing a virtual presence above those environments. Most human organizations can exist in this way, and become much more robust and flexible when not tied to a specific system.

Out there for the taking

By its very nature, most genuine innovation can't be predicted. But there are some obvious targets out there for the taking - extant problems which, if solved, would revolutionize the marketplace.

Marketing that works

To be clear: marketing works already. That's why vendors pay millions of dollars to television channels and radio stations. But as these media shrink, and as marketing money becomes more scarce (the demand from R&D is ever increasing) vendors are looking for a new definition of 'works': marketing that is welcomed, even requested, by potential customers, marketing that is not wasted on people who will not buy, marketing that is viewed as positive and helpful, not vile, crass and commercial. Product placement (been thinking of Cherry Chapstick lately?) is the new nirvana, but is still hit and miss.

Intelligent radio/television (live conversation / events)

Television and radio had show a surprising resilience in the face of the internet onslaught, and the reason for this is that they're easy. Turn it on and it will entertain you without pause until you turn it off (or until it shuts down for the night, an oddly archaic practice that still exists). Shows featuring online content - such as CNN's replaying of YouTube videos - have been, well, awful. But there's so much out there, and so much we could do for ourselves. If we had internet-enabled television or radio that programmed itself, that was personalized, that let us interact with it (in a meaningful way), imagine the future. Talk radio, for example, that is a conversation with people around the world that you find interesting.

Personal presence / personal health / personal learning

Personal heath records, personal learning environments, personal publishing and printing, personal presence: all of these are ways of imposing the personal on the technical, about making these tools about *you* instead of about them. This is essentially a combination of technologies - of smart cards and their mobile ilk, of content analysis and presentation, of connected applications, of distributed systems. From the point of view of the internet, 'you' are a concept - the one thing in the whole system that isn't actually a part of the system. How to leverage that will be the stuff of genius and innovation.

Simulation / immersion

It's easy to do simulation and immersion if you have a lot of money. I have been in flight simulators that are absolutely convincing. But they cost millions to build; nobody is likely to have one in their living room any time soon. But in the field of affordable immersive simulation is a wealth of opportunity - imagine being able to experience 3D environments from the *inside* (and not just viewed through a screen) without leaving the house. We had reading rooms and TV rooms in the past: the device that creates the Sim room of the future will make somebody rich.


The problem ith representative democracy is that your representative is often looking out for someone else's interests. Often his own. Internet technology creates the possibility for direct democracy, but this in turn requires a way of rethinking how we manage society. We want to connect people to government - but only those parts that affect them directly. We want to create mechanisms that allow people to govern themselves - but not to govern others. Collaborative and community-oriented systems for resource management and decision-making will be a fertile field in the future.

Energy nets

As energy becomes increasingly scarce, we will look not only to alternatives in power generation, but also better ways to manage transmission. We already have an energy grid, but as events have shown, the grid is unstable and liable to cascade failures. It also depends unreasonably on a small number of very large power sources, such as coal or nuclear powered generating plants, hydro dams, and the like. We want to be able to manage energy nets of the future using distributed sources - wind and solar powered, for example. Such systems would be tasked to minimize transmission load, insulate against cascades, and promote diversity of sources.

Dead tech


By the 'telephone' I don't mean voice-to-voice communications generally - people will always want to talk to each other - but about telephony in particular (and in particular, dedicated lines and switched services). The reason is simple: it is simply too much overhead to maintain an entire infrastructure premised on the possibility that any given person may require a direct audio link to any other given person. We want to use the wires for other things. And the overhead required to support switching is immense and expensive. The closest thing we'll see in the future is something like bandwidth guarantees for specific services .


Television is so close to being over we can almost taste it. Once digital television comes into households (2009 in the U.S., 2012 in Canada), the previous monopoly owned by the cable companies will be broken. Televisions will no longer be tubes, they will no longer have channels (increasingly, we'll just program numeric selections), and we will no longer watch networks (increasingly, we will watch providers - Fox, Gawker, Google, CNN). Yes, we will continue to have background audio and video displays in our room - often more than one - but we will no longer be 'glued to the tube'.


Radio is rapidly ending its life as an electronic transmission medium. Today, it is almost as common to listen to 'radio' stations online (through iTunes, for example) as it is to pick up signals from the air. As broadband becomes ubiquitous, we will more and more frequently simply pick up ambient internet and stream audio - whatever that entails. Satellite radio was the last harraugh of a medium that depended on mass broadcasts.

Print / paper

Paper is a resource-intensive industry and will become more and more expensive over time. Already, we are seeing the shut-down of pulp mills in remote regions of Canada, ostensibkly because they are "inefficient" but in realiuty because the market simply isn't there for their product. Bookstores are filling their shelves with trinkets, DVDs, toys and games. newspapers are losing subscribers in droves (mostly as they die off). The affordability of electronic combined with the wastefulness of paper makes this an easy prediction.


It has for the last few decades been cheaper to transport goods around the planet than to manufacture them where resources and labour are more expensive. Similarly, it has been cheaper to transport people from their pleasant homes in the country (or pseudo-country) than to live and work in the same location. This all changes as transportation becomes increasingly expensive. We will live and work in closer communities again, which means that systems that support local self-sustainability will be in demand. Can we build apartments that are as comfortable as homes? Can we grow our own grapefruits and coffees? Can we design specialized production systems that do not depend on cheap labour?

Stephen Downes Stephen Downes, Casselman, Canada

Copyright 2024
Last Updated: Jul 24, 2024 7:33 p.m.

Canadian Flag Creative Commons License.