Self aware computing - and why it might accelerate the end of the desktop computer

Two examples from the last few hours:

1) My Android phone wants to download an update, but won't do it now because there's only 18 per cent of the battery left (it says it wants at least 20 per cent).

2) My Windows 10 computer running Office 365 has just crashed and I've lost about 15 minutes of work since the last Autosave of a Microsoft Office document. It happened after I separately tried to use a Google Drive encryption tool called Boxcryptor.

What's happening here?

It is easy to see the commercial logic going on.

In the first example, Google has a big incentive to make sure Android works well, and with 2 billion users it can afford the budget to do that - and "working well" means avoiding a scenario where the computer might crash, such as running out of battery power partway through an update. It has strong control over the different software elements - and in the case of an operating system update, complete control

In the second example, Microsoft has little control over the customer's choice of software tools and how they might interact, and not much incentive to test every possible scenario. They don't suffer any commercial harm if a customer decides that on balance it may be better to stop using third party software on Windows.

But still - there's a lot of room for improvement. In 1996 working with Microsoft Office might mean anticipating one computer crash a day - we're better than that now but not crash free. Every crash means lost time, perhaps lost work, frustration or worse.

There is a technical discipline looking at how this can be better and it is called "self aware computing." Given the amount of PC frustration out there, surely there is a business case for it.

Self aware computing isn't necessarily about computer crashes - it can be about any system which adjusts to its environment, such as a robo-tractor-mower which can avoid an obstacle. But perhaps keeping computer systems working reliably is the most interesting market.

An important point to note about self-aware computing is that it is very hard to add in retrospect. For example you could have a software module which sits on top of all of your other desktop work, knows exactly what you are doing, does 'big data' analysis of another customer who also used Boxcryptor in the same way I did and exactly why the crash was caused, and can warn me of what I am about to do and sell useful intelligence to software companies.

But that seems unlikely to happen - there are so many different scenarios of software and updates people can be running, root causes of crashes are so hard to determine, and such a software could use enormous resources. There could be big privacy concerns if it is basically sending screenshots and detailed data about computer operation somewhere. And updating desktop software in an attempt to solve it may cause more problems not less.

Perhaps what is more likely to happen - and perhaps more desirable - is a big drive to simplify and standardise computer systems, so it is easier to understand what is going on, and the software manufacturers can much easier embed 'self awareness' right from the start - as - it seems - Google has done with Android.

Or maybe the desktop computing arena is the last place to look for self-awareness. We can already find it on our cars, where sophisticated systems are continually adjusting fuel supply and spark taking into account what they know about the engine operation - and can do it in the knowledge there won't be any pesky software encryption tools getting in the way.

Come to think of it, perhaps this could be the pathway to the end of desktop software - few office workers have an easy time of it and our tablets and phones are able to do nearly everything desktop software can do.

[thanks to Dimitris Lyras for many ideas here]