The author

Ten and the art of computer-power maintenance:
the ten €/$/£ computer.
A contemplation on values

Steven Pemberton, CWI and W3C, Amsterdam

Prelude: updates on my previous talks

2008: Why you should have a website

I introduced RDFa as a new technology: Now adopted by Yahoo, Google, US and UK governments, Drupal, just to name a few.

2009: Never is a long time

LED lamps: I quoted "Affordable 60-watt-equivalent LED in 2 years". In fact Albert Heijn has been selling them since October (if you consider €17 affordable).

Linux: I quoted "Linux will outship Windows on new PCs in 2009". Clearly he was premature...

USB stick prices:

USB Stick prices

USB Stick price developmentIn fact, since then prices rose slightly, and only now are dropping below the price of a year ago (32G now under €40).

Ten

We all know Moore's law. Things double at regular intervals.

It is an exponential growth.

Note that a doubling per 2 years is the same as a 10 fold increase every 6 and a bit years; it is the doubling that is important, less the period.

Exponential, 20 iterations

Graph of 2^x

People talk about passing the 'knee' of an exponential growth. But look at the bit up to 10:

Scale - 10 iterations

Graph of 2^x from 1 to 10This too has a 'knee' even though it's the same graph. And look at the graph to 40 iterations. It too has a knee, but much later than the graph to 20 iterations:

Scale, 40 iterations

2^x from 1 to 40Note how there seems to be nearly no action before iteration 26. Now look back at the 20 iterations graph. This is just a problem of scaling: there is no 'knee' in exponential growth, it is just an artifact of the particular scaling you use. That's why we use logarithmic scales:

Logarithmic scale

Using Logarithmic scaleWith a logarithmic scale, the vertical axis rather than going 1,2,3,... goes 1, 10, 100, 1000, ... On this scale, a true exponential is always a straight line.

Example of exponential growth: Laptop speeds

Laptop speeds

Exponential Bandwidth Increase

Bandwidth on a log scale

What exponential growth really means to you and me

Often people don't understand the true effects of exponential growth.

A BBC reporter recently: "Your current PC is more powerful than the computer they had on board the first flight to the moon". Right, but oh so wrong.

Take a piece of paper, divide it in two, and write this year's date in one half:

Paper

2010

Now divide the other half in two vertically, and write the date 18 months ago in one half:

Paper

2010
2008

Now divide the remaining space in half, and write the date 18 months earlier (or in other words 3 years ago) in one half:

Paper

2010
2008
2007

Repeat until your pen is thicker than the space you have to divide in two:

Paper

2010
2008
2007
2005
2004
2002
2001
1999
98
96
95
93
92
90

This demonstrates that your current computer is more powerful than all other computers you have had put together (and way more powerful than the computer they had on board the first moonshot).

Orders of magnitude

Each step up in a power of ten is called an order of magnitude change.

"An order of magnitude quantitative change is a qualitative change".

Half way between 100 and 1000 on a logarithmic scale is about 300 (316.2 to be more exact).

So we say that for an order of magnitude, 300 ≅ 100

Also, since we are only talking orders of magnitude, it doesn't really matter what unit of currency we use, so I will use ¤, which believe it or not is the generic "currency" symbol in Unicode (and Latin 1).

Disruptive technologies (slight return)

2009 was the 400th anniversary of an Englishman, Henry Hudson, working for a Dutch company, discovering the island of Manhattan. Then, and for centuries after, all trans-Atlantic shipping was done with sailing ships.

Hudson and the Halve Maen in the Hudson River

Steam ships

When steam ships were introduced they were not reliable enough to travel trans-Atlantic distances, they couldn't travel far without breaking down, and they were inclined to blow up.

An early steam ship

Niche

But steam ships were able to find a niche in lake and river transport, where the distances were short, and where they had the advantage of being able to travel against the wind and on wind-still days.

Once in the niche they could improve reliability until they were able to travel trans-Atlantic.

Once that happened, all shipping switched to steam, and all the companies producing trans-Atlantic sailing ships went out of business; not one survived into the 20th century.

Sustaining improvements

Sustaining improvements Typically, technologies improve gradually and continually over time. These are sustaining improvements. An existing technology fulfils or exceeds the market's needs.

Disruptive improvements

A disruptive technology is introduced Steam ships on the other hand are an example of a disruptive technology, a term introduced in the excellent book The Innovator's Dilemma by Clayton M. Christensen.

A disruptive technology when introduced is able to do similar things as an existing technology, but has one or more disadvantages that make it less desirable for the existing market, such as higher price, lower performance, lower reliability etc.

Niches

A disruptive technology improves to beat extant technologies If that were all, then of course no one would buy it, but a disruptive technology always has at least one advantage over existing technologies that despite its disadvantages allows it to find a niche, and therefore survive.

Whilst in this niche it is able to improve, typically more rapidly than the existing technology, until it performs equally with the existing technology, or at least sufficiently for the market needs, but with the added advantages that it had that allowed it to survive.

As a result the existing market steps over to the new technology and the old one dies (typically taking the companies producing the technology with it).

One computing example: 5¼ inch disks

It is difficult to believe that the change from 8 inch to 5¼ inch disk was a disruptive innovation, but they represented different technologies. Compare the two when 5¼ inch disks were introduced in 1981:

8 inch 5¼ inch
Capacity 60 MB 10 MB
Volume 10 litre 2.5 litre
Weight 10 kg 2.75 kg
Access time 30 µs 160 µs
Cost $3000 $2000
Cost per Mbyte $50 $200

Stages of introduction of a disruptive technology

When a disruptive technology is first introduced, it looks pathetic to the existing market. "Why would anyone want that."

Disruptive technologies are almost never introduced by an existing company, in the few cases that an existing company has seen the risk, and been successful, it has been by spinning off a new company or division.

Reminds me of: "First they ignore you, then they ridicule you, then they fight you, then you win." (Attributed to Gandhi - on a completely different subject)

The Generations of Computers

Thanks to the effects of Moore's law, each decade or less, there has been an order of magnitude increase in computing power.

This has enabled a new generation of computer to appear.

"An order of magnitude quantitative change is a qualitative change"

As a result we have used each generation of computer in a new way, not just for more of the same.

The Mainframe

From the 1950's the way computing was done was on mainframes. These were room-sized machines, shut off to the outside world, and would cost millions. They were produced by the likes of IBM, Univac, Burroughs, Control Data, NCR, Honeywell, and others.

They were so expensive that many companies would lease computers rather than buy them.

They often came with 'free' programmers into the bargain.

Burroughs

Honeywell

Univac ad from the 50's

Using a mainframe

Let me describe how I had to use a mainframe in my first programming job.

I would think out my programs on pieces of paper, then write them out, before copying them onto specially printed sheets

Coding sheet

...

Punch card machineWhich sheets I would give to a punch-card operator to type up, who would pass it on to the verifier operator to check it.

Why use two extra people? Because it was cheaper than running an incorrect program on the computer!

...

Punched card deck

When ready I would take my pack of cards and walk over to the computer centre where the mainframe computer was kept. (Of the order of 100k bytes per box).

...

I would type up a special header card describing the job and the resources needed by my job, and then hand over the cards to an operator who would put them into the queue of jobs to be run.

Then I would return to my office, and work on some other problem.

...

About an hour later I could walk back to the computer centre, and if I was lucky my cards and the line-printer output from the job would be waiting for me to pick up.

Example printer output

I would then quickly peruse the output. If there had been a syntax error, I could correct the cards involved and resubmit the job immediately. If the job had run satisfactorily, or there was a problem that needed more thought, then I would walk back to my office.

...

If I was unlucky though, my job wasn't ready. I would then look at the status of the jobs to be run from a printout that was hung on the wall at regular intervals, in order to get an impression of how much longer I would have wait. Jobs that needed a minute or less of computer time got priority; if you needed more, you often had to wait a long time before your job arrived (more than a day).

...

This was the typical life of a programmer of a mainframe computer. Compared with modern programming, you got lots of exercise and fresh air walking back and forth to the computer centre carrying heavy boxes of cards.

The Minicomputer

Starting in the 60's but picking up momentum in the 70's, a new type of computer started appearing, the minicomputer. These were more cupboard-sized machines, costing of the order of ¤100,000, and manufactured by the likes of DEC, Data General, Prime, and Wang.

pdp-11 The disadvantage of these machines were that they were slower, and had few resources (memory, disk) than the mainframes. But the advantage was that they were cheap, and you could have them in the lab. You still had to share, but it was nearby, and you got instant turnround once you were on it.

Why are so many names unknown now?

Univac, Burroughs, Control Data, NCR, Honeywell, DEC, Data General, Prime, and Wang. Why aren't they around any more?

This is the result of the typical response to disruptive technologies.

An existing company with an existing group of customers is unlikely to see the new technology as tempting: it is under-powered, and worse, has lower profit margins.

It usually takes a new company to come along (such as DEC with the mini computer) to develop a new technology. Very rarely does an existing company successfully make the transition to the new technology. That's why it is called 'disruptive'.

The Workstation

SGI workstationStarting in the 70's but picking up momentum in the 80's came the Workstation, which could go on the desktop of the programmer in the office, and were known as 3M machines:

They were made by the likes of Sun, Xerox, Perq, Lisp Machines. Now at last programmers had machines of their own.

The PC

Starting in the late 70's, came the PC, the first computer to make its way into the home (and the briefcase). It cost of the order of ¤1000, and was made by the likes of IBM, Apple, Compaq, Atari, Commodore.

Apple ][

And so

Since we have seen

¤1000000 mainframes

¤100000 minis

¤10000 workstations

¤1000 PCs

We really should expect the emergence of the ¤100 machine

And indeed they have emerged

¤100 computer

Netbooks

Netbooks too are of the order of ¤100.

asus eee

Disruptive?

Ridicule: Michael Dell: Netbooks go sour after 36 hours "We see a fair amount of customers not really being that satisfied with the smaller screen and the lower performance - unless it's like a secondary machine or it's a very first machine and the expectations are low," he said. "But as a replacement machine for an experienced user, it's not what we'd recommend. It's not a good experience, and we don't see users very happy with those."

Real usage: they have found a niche, so PC manufacturers beware!

Expectation: netbooks/tops will grow to destroy PCs (you might think they are the same, just like 5¼ inch disks and 3½ inch disks seemed the same.)

Expectation: some PC companies will die (Toshiba apparently already going - to Asus)

What we have seen

Overview

Why did the ¤100 computer take so long to arrive?

When the XBox was first introduced (2000, $300), many people wondered why we couldn't have computers at that price.Original xbox

Why did the ¤100 computer take so long to arrive?

And we did have a £100 computer 20 years before that in 1980:

Sinclair ZX80

Usage of the ¤100 computer

Carry round, plug into larger screens and keyboards.

No longer one computer per family: really do need new operating systems

Use NAS for family storage

Bandwidth is doubling per year (remember floppies were 64kB/s): use NAS or cloud uniquely. Google docs, Flickr, Facebook.

Google OS is coming

= Return of the thin client

So are we done?

Should we be thinking of a soon-to-appear ¤10 computer, and if so, how will we use it?

Calculators

My first pocket calculator cost (my employer) of the order of ¤100 (about 1/10th of my annual salary at the time). Now such a calculator is essentially free (I have a drawer-full of such things that have been given to me).

1972 Pocket calculator

Usage

Typically people expect that w will use new technologies in the same way we use existing ones.

Steam engines: One engine, lots of pulleys to distribute the power over the factory.

Assumed the same would happen with electric engines: one engine in the huse with pulleys taking the power to where you needed it.

Vacuum cleaner point in every room.

Same with mainframes. Assumed 5 would be enough. Why would people want personal computers? They don't need to do payrolls!

Usage

Uses around the home. How many CPUs do you have in your home now?

One computer per light fitting? Why not?

Might we plug it in like a USB memory stick? I doubt it.

Can we already see the coming of the ¤10 computer?

Mp3/Mp4 video player with colour screen: under €30

Keyring picture frame: €10

Keyring picture frame: €5

These examples all have a screen, a USB port, and can do reasonable amounts of processing.

Conclusions

An order of magnitude quantitative change is a qualitative change.

We have seen 5 different generations of computers since the 50's. We should anticipate the next one.

And how about the ¤1 computer?