Everything for nothing part 2

I seemed to have caused some heat a few days ago when I suggested the future of corporate IT was “everything for nothing”. My point was that commoditisation of technology inputs would reduce their price to a point where they would, effectively, be “too cheap to meter”.

Since then, many people have written to me privately saying they “don’t get” the argument I’m making, though most have been polite enough to stop short of telling me I’m wrong.

So I thought I’d better explain myself a bit better, perhaps with the aid of some of the public comments on this blog.

Let me start with the basic refutation of free from Anon, who reduces the argument to economics:

…until food, clothing and shelter are free how people who make the free stuff for others to consume can make a living. Nothing is free to make for others to use. The Internet is not 100% free as someone needs to be paid to keep it up and running and they use this payment to buy food, clothing and shelter.

OK, so I admit at this point some dramatic licence when describing the price as “nothing”. Quite clearly, someone is going to pay for stuff, particularly the stuff that uses atoms. Perhaps I’d have been better off using the “too cheap to meter” phrase instead.

Now this is a world we’re already in. I buy an internet connection today, and pay about ten pounds a month. Every year or two, the width of my pipe has doubled without much in the way of a price increase. The trend is obvious, and is the same for the computers I’m using, and everything else in my life that depends on IT. My bandwidth, anyway, is too cheap to meter. So is all the rest of the technology around me.

I mean, there is a CPU in my light switch now. It knows how to do the dimming and switch-on things when I point a remote at it. Those are cycles that are too cheap to meter.

But Thomas Barker, regular commenter here says this:

Focusing on CPU cycles is a red herring. They aren't the bottle-neck in business systems. Most systems spend all their time on I/O, so if anything, you'd expect massive consolidation *within* datacentres to economise on disk activity and storage. Not processing being dynamically relocated across continents. The bandwidth to push the data around is more expensive. CPUs are cheap.

Now, this may be true today. But the inexorable price decline in bandwidth on public networks is irrefutable. Its easy to see when bandwidth will be cheaper than cycles.

Anyway, Thomas has missed the point, I think, when it comes to cycles. Its not the chips which are going to be expensive, its the power. There is going to be an energy shock sooner or later, just like the oil shock on 1973. That had the result of hugely inflating the price of crude, with the result that everyone suddenly had to be concerned about fuel efficiency.

It is not hard to imagine a carbon tax on datacentres in the next decade. What would that do to the price of your processing?

Most countries are inefficient producers of cycles. When you can’t get enough power into your datacentre no matter what you do, you’re going to be forced into running your workloads virtually elsewhere. I’ll not make any predictions about storage, except to say that if you have virtual workloads moving around, I’d expect the storage to move with it. Thomas correctly identifies that having the data with the workload is usually a good thing.

But on to the desktop. Internally, I got an email from someone that reminded me that “the cost of a desktop isn’t the hardware, its the support to make it all go”. I paraphrase, but the point, really, was that bodies have to sit around to fix things that stop working.

The fact of that matter, though, is that there is a growing population of people in enterprises who are able to fix things themselves, and who, actually, have better skills at doing so than the corporate helpdesk. When you let them get on with managing their own tools, they’ll be much more productive that that big lumbering corporate helpdesk would have let them be.

That’s not everyone, today. But it is a growing body of people, and sooner or later, its going to be a majority.

I want to close with another point from Thomas’s comment:

As for people hacking their own business programs, it can help, it is how most useful things start, but scaling, or integrating, stuff like that... eeek!

I know you all know my argument here is going to be obvious: Open Source does all of these things perfectly well. Software that evolves from the bottom up works, even in a business context. Having your end users do it rather than coders is a question of waiting for the right tools, not any lack of motivation or capability.

I cannot imagine that end-user computing tools a decade from now won’t routinely be capable of doing big systems.

I have one final point to make. Most of the communication I got on this post came from IT professionals, all of whom said that none of this could happen for various reasons. The discussion I got from non-IT colleagues, though, was along the lines of “but isn’t this obvious?”

What conclusions would you draw from that? Here are mine: big IT will be over in the next decade or so, and all those professionals who manage the commodity inputs had better get themselves up the foodchain quick smart. Those that fail to recognise that they’re not that special any more will be out.

And if you sell the commodity inputs, be prepared for your nice premiums to crash. Our needs aren’t so special that you can charge us over the odds. You too, will be out unless you can dream up something that makes you special again.

Comments

[Colm O'Connor]

>I know you all know my argument here is
>going to be obvious: Open Source does all of
>these things perfectly well. Software that
>evolves from the bottom up works, even in a
>business context. Having your end users do
>it rather than coders is a question of
>waiting for the right tools, not any lack of
>motivation or capability.

Generic open source often does all of these things well. If you want a basic small business accounting package, I've no doubt that you'd be able to get a decent, scalable, integrable one for free. Even now.

However, if your users want something specific (even accounting that is specific), they'd need some sort of expertise in data modeling, which it both unintuitive and non-obvious. Learning what is a sensible key to use is something that even trips up experts, let alone end users who just want a tool to get something done.

I couldn't count the number of systems I've run across designed by the end users that have awful data consistency issues. This is before you even get to the problem of data transformations, import/export, encryption, etc.

I can imagine a world where these things are in principle handled by business analysts, but if you teach them all the skills they'd need to be able to do all this stuff, AND handle all of the edge cases, exceptions, issues that arise they'd become skilled IT people.

...and then you're back to square one.

[Henry law]

> here is a growing
> population of people
> in enterprises who
> are able to fix things
> themselves, and who,
> actually, have better
> skills at doing so
> than the corporate
> helpdesk.

I don't deny that: I'm one of those people. The problem is that many others will lack the inclination or the intellectual horsepower to do that. What they do is feed off the people in their community who are good at it. Result, Bert Pheggs becomes the departmental help person; the work that he was specifically hired to do suffers, and he's paid an order of magnitude more than an offshore help desk person.

[Colm O'Connor]

Also, you will always need IT people for security. The economics of security don't work in the same way - it's a continual arms race, not a relentless commoditization drive to the bottom.

[Thomas Barker]

I'll write a blog post this weekend :)

[Peter Johnson]

On the initial reading of James' original article I was a bit prickly. No one likes to think that they'll soon be obsolete. Then I thought about it quite a bit and realised that he may be right. However if one gives the subject some detailed analysis it's clear this isn't the only future. It's a values thing, and there are dangers (see Colm O'Conner's comments above) inherent in the future that James outlines.

Anyway James' piece isnpired a blog entry of my own so it's all good: http://bit.ly/PtTGE

[David]

"there is a growing population of people in enterprises who are able to fix things themselves, and who, actually, have better skills at doing so than the corporate helpdesk"

Maintenance of a modern computer desktop is not easy, especially computer desktops which are completely unlimited. Almost any IT people with families will attest to the fact that they are frequently called in to help with issues because "you know about computers don't you?". Also most people who "know a bit about computers", know about computers in the singular or low numbers sense. They don't know all the many and varied wrinkles that come from running an estate of thousands, tens of thousands or more computers and keeping some kind of harmony.

This is why we have IT specialists. They have the knowledge and experience of what to do in a large scale environment so that the users don't have to. Now arguments could be had about how much freedom the IT specialists should give their users, but I don't think the world of IT is at a place yet where the users can do it all on their own without causing chaos. I'm not sure I see us being there too soon either, because these are not easy problems to solve and one size does not fit all.

Re end-user computing
"Open Source does all of these things perfectly well"

Well, yes and no. Yes, open source is often great at getting things done and can be very effective. Open source email servers and databases are really good examples of (by and large) well programmed, scalable, near-enterprise level software that works really well. But can *you* configure it? (genuine question). Open source coders are frequently excellent at solving the problem and making the program highly efficient, but human interaction experts they frequently are not. The email server "Sendmail", involved in 65% of global communications is cryptic in the extreme, horrendous to configure and very easy to configure badly. Even "IT Experts" struggle with it.

Apple is a great example of this. Most of the basis of the OS X operating system is open source free software, so why do so many people spend a not insignificant amount of money on apple computers when they could so easily run Linux on something much cheaper? I believe that much of it is the extra added value of "ease of use" for the consumer. The end user doesn't want to care about how to configure his computer, and he's often prepared to pay so as not to worry about how to configure the intricacies of their computer. Even I do and I am an "IT Guy" who has used Linux and FreeBSD both on the server and desktop.

Now this is fine, it's the job of IT experts to know how to configure the tools of the trade, but surely this is not a great model to suggest for getting unqualified users across the estate to follow when solving all the IT problems of a big organisation.

For smaller scale deployments then I agree that end user computing can be an excellent model. Large scale IT projects however are traditionally expensive because they are difficult to do. They feature clients who have no idea what they want and change their requirements frequently, a changing climate of regulations, guidelines and best practices, matching enterprise software featuresets (resilience, backups, scalability, performance, integration with other products etc). And all of this ignores the confidence one should have that the end product actually does what it's supposed to. None of this is to do with the right tools, most of it is to do with how you use those tools correctly and I'm not sure I'd expect the user to know or want to know.

Should the CEO/CFO of a large organisation be happy that the year end accounts for their large multinational, multi-unit company are computed by steve and terry, the junior accountants? If they should make a mistake and the numbers are wrong, who is responsible? Who would the CFO blame? Who would the shareholders blame?

If your mortgage was with a bank who calculated your payments on an excel spreadsheet designed by kim at the swinton branch, with help from colin in peebles.. Would you be happy? Would the bank be happy to go in front of the FSA or court to prove that the payments were calculated correctly?

And lets not forget that writing such systems takes a great deal of time, even if you ignore all those time consuming things like testing, documentation and assurance. This is time that the user's manager might reasonably consider ought to be spent doing their actual job.

I would say I'm not inherently against end-user computing, but I think that it's not suitable for everything by a long shot.

"I have one final point to make. Most of the communication I got on this post came from IT professionals, all of whom said that none of this could happen for various reasons. The discussion I got from non-IT colleagues, though, was along the lines of “but isn’t this obvious?” What conclusions would you draw from that? "

I find it interesting that when faced with feedback on your IT focused article by both non IT folk, and by highly experienced and qualified IT people whose job it is to know how IT works and how best to make efficient use of it, you think that the non-IT folk are bound to be correct. Is there not a chance that the IT folk do know what they are talking about?

Personally, I think that end-user computing will grow slowly but surely, I think that many commodity IT services will become a lot cheaper or bundled, but that there is always a place for big IT, because people want big IT. People want to know that qualified people have designed the systems that run their lives, they want to know that the system is considered "good" and that someone is responsible for it.

David

[James Gardner]

David,

I am also a "highly qualified" IT person, and I agree with my business colleagues. IT folk do, without a doubt, no what they are talking about. At least, when they are talking about the things we do right now.

But if there is one thing I've noted, it is the pace of democratisation of technology is speeding up, not slowing down. I don't think it unreasonable that in ten years, we could easily be in a situation where our users build our systems for us.

This is an opportunity, not a threat, in my view.

[Thomas Barker]

Here's a sort of response.

"The cost of silicon has collapsed; power usage is under control; bandwidth isn't an issue if you're sensible. "Enterprise" is a misnomer, most it is barely distinguished from what's on sale at Dixons. Within a few years, IT in well run companies will focus almost solely on improving the use of people. CTOs of the future will spend their lives trying to cut phone support and replacement call out costs.

Presumptions that businesses need ever higher capital expenditure on processing power are now demostratably false. Developments in technology have far outstripped the true needs of organisations. Servers capacity is doubling every two years; the data any business needs to processes is not. Given that creating a business data point usually requires a real-world event, it is not physically possible for general business needs to keep up with technology................................"

..........http://thomasbarker.com/09/11/big-condensation

[Ken Maier]

I may be a bit late with a few comments- however as the saying goes..

Many years ago I attended what turned out to be one of the most interesting workshops of my career sponsored at the time by Peat Marwick (anyone old enough to remember them?- KPMG now) Anyway, the moderator Bruce Stewart began the session by asking the following very simple question: " What Is A Sale"

Now for anyone in business, sales, or politics(nowadays)the answer comes pretty naturally.. however when we peel the onion a little - I see an interesting connection to "Everything for Free"

After asking each of the participants of the workshop for the opinion of "what is a sale" - and some pretty interesting responses.. Bruce informed the group of the following: " A sale can only take place when the "percieved" value of what we are buying is of greater value than what we are paying"

With the exception of taxes- think about anything you have bought in the last 48 hours and why you bought it and why you paid the price you did.

As more and more products and services are commoditized into "better sameness" does this mean that the price necessarily has to go down or be free in order to be valuable? Or do providers keep increasing the "percieved" value hoping that we will continue to pay the same amount. Examples of this exist everywhere in the product and services we consume. Even products and services that are "free" can cost too much and I expect that ultimately free will come with a unexpected and decidely higher price tag somewhere else..