The Real-World State of Windows Use 374
snydeq writes "Performance and metrics researcher Devil Mountain Software has released an array of real-world Windows use data as compiled by its exo.performance.network, a community-based monitoring tool that receives real-time data from about 10,000 PCs throughout the world. Tracking users' specific configurations, as well as the applications they actually use, the tool provides insights into real-world Windows use, including browser share, multicore adoption, service pack adoption, and which anti-virus, productivity, and media software are most prevalent among Windows users. Of note are the following conclusions: two years after Vista's release, not even 30 percent of PCs actually run it; OpenOffice.org is making inroads into the Microsoft Office user base; and despite the rise of Firefox, Internet Explorer remains the standard option for inside-the-firewall apps."
Spyware (Score:3, Insightful)
The Windows Sentinel app:
When they sell your info it's spyware
When they post it on slashdot it's a community-based monitoring tool
Re:Browser use isn't exclusive (Score:1, Insightful)
And it's also why I'm unable to use netflix's watch on demand feature on Linux :-(
I'm enough of a die-hard to use Linux anyway, but that kind of thing is enough to keep most folks from using it. It doesn't "just do" a bunch of what they want to have work.
Representative? (Score:5, Insightful)
This is only representative of the 10,000 PCs running this software downloaded from InfoWorld, it would appear. This doesnâ(TM)t sound like it has anything to do with the âoereal worldâ unless you think that the subset of Infoworldâ(TM)s readers who would download this software are somehow representative of the broader Windows population.
Need to retake to Introduction to Statistics .... (Score:5, Insightful)
Of note is the fact that, two years after Vista's release, not even 30 percent of PCs actually run it
No, not even 30% of the subset of PCs with this performance-monitoring software run it. In order to claim that not even 30% of PCs run Vista, you would need to establish that the sampling method is not biased, which is a pretty implausible claim.
It would not surprise me if the subset of technically savvy PC users are biased towards XP and that subset of "Windows is what comes on the computer from the store" have whatever the store put on it.
What the hell? (Score:5, Insightful)
If you have to sign up to be a part of the data gathering, it is NOT real-world usage, as the other billions of us out there haven't signed up.
And another thing.
The summary quotes a number of 10,000 sampled machines, yet the number in the first link says 20,000.
Which is it, boys?
A +/- variation of 50% in something as simple as the number of machines sampled leads me to believe there more then likely other errors.
Re:Spyware (Score:5, Insightful)
Re:Total Survey Numbers (Score:2, Insightful)
Or in two words: sampling bias.
Legacy Software (Score:5, Insightful)
10,000 PCs is a small sample size, try a few million. You might have a sampling error there if they are not randomly picked.
One reason why Windows Vista has not caught on is that older hardware won't run it, like my Father's Pentium 4, 512M, Windows XP Home System, it is not listed as Vista compatible and fails the Vista upgrade check. The memory cannot be upgraded to more than 512M due to motherboard limitations, and the video is not Aero compatible and there is no video slot to upgrade it. I doubt it will run Windows 7 either. Trying to force a Windows Vista install on it will mean that it will run slowly (512M is the minimum I know, but with that size memory Vista runs slow) and some features would be disabled.
My own laptop a Compaq Presario F700 series came with Windows Vista Home Premium on it, but it caused random lockups that Microsoft blamed on Compaq, and Compaq blamed on Microsoft, and after going in circles trying to get help I downgraded it to an OEM copy of Windows XP Pro that works without any problems at all. But I have a Windows 7 Pro upgrade coming in October to try it out. Hoping that if Windows 7 stinks as much as Vista did, that I can go back to XP Pro. On the other hand Fedora 11 works with the wireless card and it would make a good Linux based laptop when XP retires and there is no more updates for it. I just wish that Visual BASIC 2008/2005 works with WINE, because currently it does not, and I need to keep my VB skills up to date for possible jobs or contract work. Something about needing the BITS service installed to install the software. Otherwise the outdated and ancient Visual BASIC 6 works in WINE, but hardly anyone calls for VB 6.0 skills anymore.
Re:Browser use isn't exclusive (Score:3, Insightful)
Re:Inflated Numbers (Score:3, Insightful)
This is a fav argument as always - the problem is that when you look at OS share collected by online data aggregators like NetApps it seems someone is actually connecting these mythical warehoused copies of Vista to the internet.
Personally, I think it's amazing Microsoft found a way to make unsold boxed DVDs of Vista to the internet. They might struggle to make Aero run on older hardware, but they're brilliant at wireless networking and power management.
By the way, 30% of roughly a billion PCs in the world... I'd like to have me a "complete failure" like that one every two or three decades.
Re:Legacy Software (Score:3, Insightful)
10,000 is a very large sample size, and adequate for almost anything. Vista's not *so* rare that it won't show up on a sample of 10,000.
However your point about random sampling is valid, although it would be just as big a problem with a sample of 10 million. This is a self-selected sample, so is highly likely to suffer from this a great deal.
Re:Legacy Software (Score:1, Insightful)
10,000 PCs is a small sample size
No, it's not. That's a huge sample size. Lots of legitimate statistical studies use about 1,000 samples, to generate confidence near 95%.
If you want to pick holes at the data, look at the selection bias. (Compare it to Debian's popularity contest application, which suffers from self-selection bias, but that bias is measurably small. Obviously, every Debian user is self-selecting. On the other hand, there is little self-selection bias for the popularity contest, since every user is presented the option to turn it on or off)
Re:Browser use isn't exclusive (Score:3, Insightful)
Dang, now he has to admit he actually likes the little rascal
Re:Need to retake to Introduction to Statistics .. (Score:1, Insightful)
And there are 3 billion PCs in this planet, so 10,000 PCs ran by geeks that have some sort of creepy exo.esqueleton.spyware.thing installed are a REAL, TRUTHFUL, TRUSTFUL, RELIABLE data source. /. oblivion keeps amazing me, year after year...
Re:Need to retake to Introduction to Statistics .. (Score:2, Insightful)
Re:Inside the (Corp.) Firewall no one can ... (Score:4, Insightful)
booting a separate os during a break to read a web page is hardly a productive use of one's personal time.
Your company restricts flash drives but allows you to boot from an optical drive?
Emperical evidence of bundling (Score:1, Insightful)
And here we have emperical evidence that Microsoft's bundling of IE does hurt the competition. OpenOffice can get a foothold on Windows becuase its competitor costs money, but Firefox can't because its competitor is free, and is built into every copy of Windows.
Re:Still, 10,000? (Score:2, Insightful)
You clearly don't understand statistics. 10,000 samples is a very large study.
If there's a problem with the data (and there probably is), it is because of selection bias.
Re:Representative? (Score:2, Insightful)
Totally agree. And even whether it's 20K PCs, as the linked article says, I'd still not represent anything...
You don't understand statistics, do you?
You have 170,000+ computers. Great. That is not a random sample. A random sample of 10,000 computers is enough to generate a confidence greater than 95%. It doesn't matter how many computers there are in total. Whether it is 1 or 100 billion or 100 million billions. 10,000 randomly chosen samples give you more than 95% certainty.
There's a reason to doubt this data, but sample size is not it.
and as usual the StatisticAreFlawed crowd is here (Score:3, Insightful)
and they aren't right.
They aren't even wrong.
Re:What the hell? (Score:2, Insightful)
Re:Inside the (Corp.) Firewall no one can ... (Score:4, Insightful)
You mentioned Ubuntu.
Even though you're using it in a completely counterproductive manner, it will definitely get modded +5 insightful.
Re:Windows as a Real World State? (Score:2, Insightful)
Re:Need to retake to Introduction to Statistics .. (Score:2, Insightful)
Re:Windows as a Real World State? (Score:3, Insightful)
We really need a -1 I want to kick you in the throat
Please refrain from reckless use of analogies.
Or a -1 Stop Requesting New Negative Mods.
Inside the firewall (Score:4, Insightful)
Firewalls are pretty good to avoid things from outside getting in by themselves. But once you put an agent there that opens the door to things from outside (probably the most used vector right now) it turns the firewall meaningless. And if well you can put things that do a virus scan of what is coming, its not easy to detect 0-day attacks, or targetted trojans, or the js/activex/dynamic html/whatever attack of the day.
Re:Inside the (Corp.) Firewall no one can ... (Score:5, Insightful)
IE6 supports many extensions and has many bugs (at least things that work different from standards but consistent in IE6), that were at the time nice for programmers and were used really a lot. Thus many internal software used in companies needs IE6 as front-end.
Many of those behaviour issues have changed in later releases, also many of the hooks have been removed for being unsafe amongst other reasons. Thus your nice corporate app doesn't work anymore in later IE releases.
Then you can go "oh but then just fix it!" - why fix it when it isn't broke? IE6 still works, right? It will likely require a large if not total re-write of the app, which is very very expensive and needs to go through all the testing that the current app has gone through over the years. It is also very well possible that the original developers do not work at the company any more (IE6 is old enough for that), so even more likely that development has to start from scratch, including getting a list of current features of your corporate app.
Blame the people originally choosing to lock themselves into IE, but you can not blame the current people from insisting to continue using IE.
What they should do of course is install IE7/8 or FF or Opera or any other modern browser as well, and restrict IE6 for those internal sites only. Still using IE6 on the open Internet is too dangerous, and more and more web sites will not support its quirks any more either.
Re:Legacy Software (Score:5, Insightful)
The size of the sample is fine, the method for picking the same is the main issue. 10,000 PC's is well over enough to get a +/- 2% error if the sample is random.
Re:Inside the firewall (Score:4, Insightful)
> Forcing the use of IE in particular for that application, as in corporate networks, is like mandating to use belt only in very slow cars.
Many companies use Internet Explorer because they want the tight integration with other Microsoft products (like Sharepoint or Office). It takes very little effort to setup a Sharepoint intranet, where people can post Excel documents and generate KPIs and dashboards and whatever the business needs to move forward. Other big software companies also have that kind of stuff, like IBM Lotus Notes or Novell Groupwise. But just like Microsoft, it's a lock-in process. Setting up that kind of environment with other software, like those open-source PHP CMS, will require a lot of work, and quite possibly, more skilled staff and a training program for users.
Also many companies use Internet Explorer because it's already built-in and it would cost more to support more than one browser. Internet Explorer is already paid for, and usually people can get things done with it, so it's a hard sale to bring in Firefox or another browser.
> Firewalls are pretty good to avoid things from outside getting in by themselves. But once you put an agent there that opens the door to things from outside (probably the most used vector right now) it turns the firewall meaningless.
Any decent firewall can have rules for both inbound and outbound traffic. Also, decent firewalls usually have DPI or other smart technologies that can really raise flags when something goes wrong.
> [...] forcing to have an agent there that is vulnerable even to bad breath (usually the enforced version is an old one, wont be very surprised if a good percent of those inside the firewall browsers are IE6) sounds almost criminal.
In my experience, companies that have a policy about the allowed version of Internet Explorer usually have a very efficient change management strategy and suffer very little downtime because of software problems. I have a Fortune 500 client that started rolling out Windows XP last year only; they have no downtime at all and the business is doing great. It's not cool or edgy to work in such an environment (even Flash is not supported), but they are not in the cool or edgy business.
Re:Emperical evidence of bundling (Score:5, Insightful)
And here we have emperical evidence that Microsoft's bundling of IE does hurt the competition. OpenOffice can get a foothold on Windows becuase its competitor costs money, but Firefox can't because its competitor is free, and is built into every copy of Windows.
Firefox has far more penetration into the Windows market than OpenOffice so what you said makes absolutely no sense.
Re:Spyware (Score:5, Insightful)
Yes, that is the way the world works, correctly. Just like the difference between you giving some guy on the street money, and the same one stealing your wallet. One is called robbery, one isn't.
Complicated world we live in, isn't it?
Re:Need to retake to Introduction to Statistics .. (Score:3, Insightful)
Re:Inside the (Corp.) Firewall no one can ... (Score:5, Insightful)
I'm guessing that's some sort of internal agreement with your IT department or outsourced IT provider, because speaking as someone with an Enterprise license agreement and a support contract with software support through IBM, there's nothing binding me or my users to a particular technology stack.
That said, we end up sticking with IE because of obsolete external applications. ADI Time's web site is an excellent example of a site that only runs in IE using Compatibility Mode, and since we use that for leave management at the insistence of HR and a long-running contract, we have to keep IE around. Personally, I'd like to move to Firefox or Chrome; not being subject to our normal software installation rules, I actually use Chrome as my primary browser. We don't install it and offer it as a choice for users because our users are, sadly, not the sort of people who'd deal well with being told, "Well, if it doesn't work in one browser, try the other one." Some of them don't grok that there is such a thing as a "web browser" and need to be told to open "the Internet" (i.e., Internet Explorer) instead.
I think the major limiters to F/OSS adoption in the corporation are a relative lack of mature management tools, the increased support costs due to user ineptitude and limited support resources, and general corporate inertia. Specific contracts that bind a company to a particular technology stack probably fall into the category of "more common than you'd expect, but less common than you'd think," and it's certainly not a Microsoft-specific thing -- and shame on the CIO who signs such a contract, or buys a piece of software that isn't cross-platform compatible.
Not to start a language holy war or anything, but we use IIS and the .NET Framework by choice, not because we're forced to do so. :)
Re:Need to retake to Introduction to Statistics .. (Score:3, Insightful)
Irrelevant.
Also irrelevant.
Well.... not to offend any Linux users, or Mac users, but those operating systems are entirely irrelevant.
For business, I only use Linux. Specifically, CentOS 5 and Ubuntu. Personally, I use Linux on all my machines. When required, I have XP virtual machines I start up on demand and then destroy. I have a single machine that I can dual boot for more demanding games, but that has been extremely rare since I hardly play games anymore and most of them work just fine virtually.
Personally, I abhor Microsoft for so many justifiable reasons.
Now that full disclosure has been performed, Linux and Mac just don't even appear on the radar. Just about every single workstation and machine (99.999%) I come across professionally and personally, is running a Microsoft operating system. When I get into the data center there are even about 1/4 to a 1/3 running a Microsoft operating system as well.
The only company Microsoft is really competing with at the moment, or at least during the scope of this argument, was itself.
That being said, Vista really is a total flop. I have downgraded more Vista machines due to instability, driver issues, compatibility issues, etc. then I care to mention. Adoption rates were horrible. I would not even install it virtually on any machine I had. Not because, I dislike Microsoft. Simply because I found the operating system unusable in the long run and it's problems extremely unpleasant.
When businesses refuse to upgrade to Vista and keep XP machines plodding along, and consumers keep asking for downgrades continually, it cannot be considered anything but a failure.
Vista was a repeat of Windows MilleniSuck(tm). Everyone knows this. Not just IT people either.
Having tried Windows 7 out, I can honestly say more than ever Vista was a failure in every way. Let's see their statistics in another 12-18 months. I am betting on Windows 7 having many times the adoption rates of Vista.
P.S - The biggest problem with this discussion is that to some, to some people, saying anything about Vista means you are anti-MS. Not true. I am just anti-Vista.
Re:Inside the (Corp.) Firewall no one can ... (Score:2, Insightful)
You should have started with "I know I'm probably going to get modded down for this, but what the hell. I've got lot's of Karma."
That, combined with your whinging about a perceived pro-Ubuntu bias on Slashdot, would have guaranteed you at least a couple of Interesting mods.
Re:Representative? (Score:0, Insightful)
By comparing numbers, I merely pointed out that no matter the size, if you have a closed or partly closed environment, you would end up having inconsistent or misleading statistics. So we kind of agree :)
Fair enough. Other people were making "the point" that 10,000 is a "very small" sample. I thought that was the point you were making, as well. As I said, "There's a reason to doubt this data, but sample size isn't it".
And yes, I understand your point, if you have a random sample of over 10,000 units, confidence is great. But what I am questioning here is: how do you make sure that's a random sample?
You have to understand the methodology by which the data was collected. This notion of randomness varies by context. Clearly, there's no point in picking any of the billions of people that don't own or use a computer, in a study about computer use (for a trivial example). Limiting the domain to the world of computer users introduces selection bias, but it is "acceptable" since we're (presumably) only interested in computer users. On the other hand, a study about 1980s computer use would very likely be unusable now, because of the changing demographics of computer use in the East. On the final hand, if we're really only interested in American computer users (or just Western users), a study from the 1980s might be fine. There is a "hierarchy of randomness" here (which can be formalized in terms of conditional probabilities/Bayes' theorem).
My guess is that if your company is like many others, and locks down computers tightly enough that no software unapproved software can be installed, then the InfoWorld readers are a better approximation of the general public than your company. It probably overestimates the use of "exotic" or obscure software, whereas your company would underestimate it. You are right that I don't know that.
Re:Need to retake to Introduction to Statistics .. (Score:3, Insightful)
I am saying that Linux and Mac are irrelevant to the discussion of whether or not Vista is a failure. Vista is not really competing with them. It was always competing with XP.
New computer sales of course have Vista on them which skews the numbers a bit. You need to remember how pissed off HP and Dell were about the constant nagging to downgrade from Vista back to XP. Working professionals had unpleasant experiences and were constantly asking people such as myself to give them back their XP. Downgrading was allowed by the EULA. I have personally converted scores of Vista licenses to XP licenses.
I go by my experience here. Vista was just a pile of shit compared to XP. Plain and simple. However, I don't think it pushed people towards Mac or Linux. Linux takes quite a bit to get somebody to adopt it. I just can't see it as an operating system that is going to replace XP for regular people that are used to Microsoft. Not to mention it just simply does not have enough programs with good user experiences to become widely adopted yet. Mac is different. The problems with Vista could of forced a lot of people towards Mac, but many that I know chose XP instead.
That's why Linux and Mac are irrelevant. They just don't affect the numbers enough to draw any conclusions about Vista.
It does not matter which way you approach this. There has been a huge resistance to Vista and nothing but complaints and disappointments from those that have used it. That makes it a failure compared to other Microsoft products.
Compare XP to Vista and it will make a lot more sense.
Re:Spyware (Score:3, Insightful)
IT Pro's are statistically insignificant, and their software use pattern is extremely predictable.
The point you raised is exactly the point of the stats - average joe's. *They* are the reason web developers (used to?) spent an extra 20% time breaking their CSS to be IE6 compatible.
Re:Inside the (Corp.) Firewall no one can ... (Score:3, Insightful)
AVG the top a/v (Score:3, Insightful)
AVG leading McAfee and Norton by a significant margine. Some other "Unknown" a/v has 35% but is not avast. These are not corporate computers.
If you click on the Get More Charts link you can see the entire array. Another I found interesting was Home Premium lead among Vista uses. Again, not corporate.
RAM has 2-3GB leading so I'd think these are mostly 32-bit systems. It would be nice if that were a metric.