Anyone who uses my marketing services, please don’t take offence. I’m not talking about you.
The title refers to that age-old debate about thin clients and fat clients that ten years ago raised the blood pressure of executives throughout IT-land . Thin clients are just simple machines that allow you to access processing power, storage and applications hosted somewhere on the network. Fat clients by contrast have their own processing power, memory, storage and applications. A nice example is a radio (thin) versus an iPod (fat).
What brought this to mind was a story from The Roundup, about tiny cubes of digital storage that would enable us to carry the world’s knowledge around with us. Surely this would mean the end of the thin client? Why rely on the network when you can carry everything with you?
Of course it isn’t that simple. It is all very well being able to store all of that knowledge, but accessing it is another matter. When you have that much information, the ability to sort, search, and filter it is what becomes valuable. And it will be a long time before a Google-sized server farm can be compressed in to a pocketable device.
The reality is that the industry has settled on a compromise position. Most popular devices today are connected in some way — PDAs via WiFi, PCs via ethernet, phones via GSM etc — but they also all have significant processing power and local storage. Even my phone has a processor running at ten or twenty times the frequency of my first PC, and it has more than twenty times the storage. To me that most certainly makes it a fat client.
Over the next twenty years this obesity epidemic seems unlikely to end. Processing power will continue to increase, even though the current generation of chip technologies is starting to reach some physical boundaries that might spell the end for Moore’s law (hence the move to multiple cores, rather than increased frequencies). Storage capacities are also increasing dramatically, as noted in the article that spurred this post.
But what is increasing quicker than either of these technologies can maintain pace is the volume of data in the world, and our reliance on computers to filter and process that data.
A simple example is HDTV. Just as we are freeing up spectrum in the airwaves by moving from analogue to digital, we increase the bandwidth required to deliver a single station by dramatically increasing picture quality. More data equals more processing power required to turn that data in to something we can use.
There is also the ‘Semantic Web’ to consider, also known as ‘The Internet of Things’.
The idea of the Semantic Web is to make all of the information on the Web usable by machines rather than just people. This would make it much easier for applications to apply some ‘joined-up thinking’ — if you wanted to book an airline ticket, your computer could get data from your diary, find out your historical choice of airlines, check the timetables and prices, and bring you back a personalised selection of options. Today doing that requires either human intervention or a very specialised search engine.
But where the Semantic Web gets really interesting is when you take it out of the Web environment. If there is a standard way for ‘things’ to express and share the data they hold, they can share it with each other to improve our lives. The oldest example is the internet fridge — your fridge could talk to the RFID chip in your milk carton and find out your milk is off. It then contacts your phone via your home network to tell you to buy more on the way home.
The point is that once items like milk cartons start spitting data on to the network, the volumes of data we have to deal with grow exponentially. Milk cartons are the new thin clients — while they don’t let us access the power of the network, they at least contribute their limited knowledge to it in a way that we, or our machines, can usefully access.
Fat — even obese — clients will be required to help us digest this information. However much filtering and filing can be done by intelligence in the network (e.g. Google agents who know your profile and find what you might like from the morass of media outlets), we will always need some local intelligence to help us interpret and interface with the data we receive.
The situation we have today with increasingly powerful networked devices looks set to continue, but the thin devices have found a new role. I, for one, intend to be a thoroughly fat device — certainly by the time the Christmas break is over.