I would like to kick off a series of retrospective posts that pull in content I have previously written about Open Source, Open Standards, and in general, being open, since they will frame all my new content here later. This one is from April, 2006, and is originally located at:
http://talk.bmc.com/blogs/blog-carl/steve-carl/linux-inflection-point
You will have to view the original to get the hyperlinks, and there are many....
I did a Podcast about this here:
http://talk.bmc.com/podcasts/podcast-carl
From Dictionary.com
Inflection Point
An event that changes the way we think and act.
-Andy Grove, Founder of Intel.
Investopedia Commentary
For example, the fall of the Berlin Wall was an inflection point in global politics and the commercialization of the Internet was an inflection point in technology.
Think of it as a turning point. When a company makes a major strategic change it is said to be "at an inflection point." This profound change could be positive or negative.
There is a classic, universal graph: the data on the horizontal and vertical lines changes depending on what you are talking about, but the basic shape of the graph is the same. The first time I saw the graph, it mapped logged-on users versus end user response time. As every new user logged in, response time degraded just a bit. Then, at some point, one more user than the system could stand would log in, and response time would go asymptotic relative to the response time axis. Kinda like this one at wikipedia. The straw that broke the computer's back, so to speak. You are running along fine... then you run out of something: memory, disk bandwidth... something. Every user who logs in after that magic point makes things not just a little bit worse, but a lot worse.
Another example: One tends to think in terms of the things that you have been studying, and I have been listening non-stop to Slacker Astronomy for days trying to get caught up on everything they have posted. One of the recent shows was about star formation and supernovii. So there you are, a bunch of hydrogen floating around in space minding your own business, when some other hydrogen mass moves into your neighborhood. No big deal. But then another mass does, and yet another and before you know it there are 20 solar masses' worth and it all gets to be too much, all of them tugging at their neighbors with their own gravitational forces, so finally everything crashes together, crunching, heating... a critical mass is reached, it all heats up to blue hot, all the hydrogen starts to fuse into helium, heat and other radiation pushing against gravity, and a Blue Giant star is formed. And that is just the first major state change. A Blue Giant runs along for a few million years, torching the neighborhood, gulping up the original hydrogen and converting it to mostly helium... and then it runs out of fuel, way faster than a G-type yellow sun that sips its hydrogen. It tries to burn the helium for a while, of course, but sooner or later it all falls apart. Everything that had been held up and away from the center by the heat and the burning goes crashing in together, smashed so tight that it's not even regular matter any more. We have a big explosion, and a small black hole is formed... and the hard radiation from the explosion kills all life like us (and maybe even roaches) for a radius of 9 parsecs (almost 29 light years.) That is what you might call a major inflection point in a star's life, a "Profound Change" on the negative side. I am glad we don't have any blue supergiants nearby.
With Linux, it is nothing that dramatic. But it is also clear that to date Linux has been growing incrementally. It has not yet hit a critical mass or inflection point where its growth is suddenly exponential rather than sequential. For instance, if you look at the data on the number of web servers from Netcraft, Linux usually just creeps up a percentage here and a percentage there each new report. Or if you look at the reports for new server sales, growth is incremental rather than exponential. This report from LinuxPlanet is 6 months old but illustrates the point to some degree.
It is also not a given that Linux will ever hit that critical turn. But here are some things that have me thinking it might:
Total number of working computers on the planet
According to the "History of Computing" at the http://www.computerhope.com/ web site, in 2002 1 billion total computers had shipped since 1972. This article at the Tech-edge web site says only about half of those are still in service, but another billion or so are expected to ship by 2007: As it points out, as many PC computers would ship between 2002 and 2007 as there were from 1972 to 2002. This is speculation from 4 years ago, too, before the advent of the 100 dollar laptop. More on that in a bit.
These 2002 estimates crosscheck with this data from the Computer Industry Almanac (C-I-A) from the end of 2004. More computers will be retired, of course, but many others will go on to their next OS: Possibly a Linux test box, or maybe a place to test FreeBSD: There does not appear to be any good way to get at the number of Linux computers that grew out of old MS Windows computers. In January, 2006 C-I-A reported one billion people were using the Internet: A good sign that at least one billion probably personal computers were actively in use, more or less. It is hard to back into the real numbers for Linux usage out there as a percentage of that. W3schools web site reports 3.4 percent of their visitors use Linux (roll down past the browser data). They also show Firefox use at 24% though, so clearly some data is skewed. They say as much in the text of that page.
"Lies, Damned Lies, and Statistics". For the purposes of this discussion, I'll just say there are somewhere between 30 and 100 million computers that run Linux at least part of the time (you know... all those MacBook Pros with Parallels installed and Fedora Core 5 and other Linuxxii, Dual booters, recycled computers, those dedicated data center servers and Linux Counters noted Linux computers, etc).
It is actually more computers than that. At least 100 million... more likely many more than that, but 100 million is a nice round number. Sanity cross check: in 2003 it was reported that Sun has struck a deal for 200 million copies of Linux for that country alone. Does not even count India, and most of South America and all they have been doing with Linux, like Microbanking and the Simputer.
Back to the 100 dollar laptop and One Laptop Per Child (OLPC). Actually, the 135-declining-to-50-dollar laptop. If this thing goes, it puts a slimmed-down Linux in the hands of a bunch of kids that are just learning computers for the first time (and that article says at the rate of "100 million a year"), although Wired reports that OLPC is "working with Microsoft on a version of CE that could run the machines". It is pretty easy to slim Linux down, of course: just have a look at Distrowatch for Linux versions like DSL or Pocket Linux to name but two.
What if the computer maven ratio holds true? If you assume that this creates a new crop of Linux computer programmers at the same or similar rates as it has in every other country Linux is in, you will suddenly have a whole new generation of programmers making the devices do whatever they need them to do. Which probably creates a feedback loop to demand...
Good enough
The Linux desktop and associated applications will sooner or later hit the "good enough to get the job done" stage. Maybe it has even now: It certainly has for me. From my point of view, it has been good enough for years, and it has not stood still. I am very impressed with Fedora Core 5, which I have installed on a test system at the house, as well as a guest under Parallels (just loaded up beta 3 of Parallels 2.1: VMware had better get their product out the door or they are going to find Parallels has stolen their normal place on the desktop) on my MacBook. I might think all the bubbles of the default FC5 theme were something like the wild eyed penguin theme from Mandrake a few releases ago, but at its core it has been a solid release. VMware is temporarily broken there for now, but other than that it appears to "just work". Plug in a device, it appears on the desktop. Click on a link, it opens fast. Each release seems to get faster on the same hardware. I have not tried Evolution there yet: I'll make a note to see if that is fixed against MS Exchange 2003 yet...
I have noticed OpenOffice 2.0+ appears to be much better with Excel spreadsheets than anything in 1.0 was. I know this is still a work in progress, but it would appear that for a certain population of folks where Excel was the barrier that you might now or soon be able to work with OpenOffice as your main spreadsheet tool. And of course there is always Crossover Office.
The "Thin Client"
I dreaded typing that heading, since it often invokes the image of a green screen 3270 or vt100 terminal on your desk. It does not seem likely that people will want to give up their "fat" clients anytime soon. If nothing else, when the network is down, you can play solitaire 
I was not really thinking of those, or even of the new release of the SunRay technology (although that stuff does seem pretty cool). What I had in mind was AJAX, and things like Google Mail and the oft-hinted-at Google Office (both by the Sun / Google deal and Google's purchase of Writely.) Well-written AJAX applications work on any OS as long as they have a modern browser on them. (UPDATE ALERT: Since I posted this last night at 2:00 am CDT, Google has now released Google Calendar. Things change fast in this industry...) The client OS is no longer important to getting your job done once you have migrated your application stack to an AJAX user interface. Other factors become more interesting. For one thing, if you run a datacenter, and you need to do Disaster Recovery, it is handy to have all the data on the servers rather than spread all over the clients. AJAX apps are lovely for this, since they use the clients' cycles to make things fast, but usually store at least a copy of everything server-side. Training is no big deal either: how you work the app is the same no matter what the client OS is.
If you move to a thin client-type approach, the thing you most want to know about your client system changes from "can it run my applications" to "is it secure, stable, cheap, manageable".. and so on.
Critical Mass
When this planet has one computer per person... call it about 7 billion active personal computers, what does that look like? I am not even counting people like me that have about 20 computers either. Clearly many of these computers are low power, flexibly powered, hardened, inexpensive, reliable, and easy to work on. Can you imagine 7 billion 100 watt computers? Talk about your global warming! The 2 watt per unit goal of OLPC seems a bit more reasonable in that context.
At what point do all the requirements and technology changes add up, and take Linux (or... some other OS: FreeBSD anyone? Open Solaris? Nah... this is a Linux column) to the next level.. take it from the incremental growth to the exponential one?
Whatever happens, it appears we'll see relatively soon. If it took 25 years to get out the first billion personal computers, and 5 years to get to our second billion, even counting system retirements, something will be hitting critical mass soon.
