Slashdot Mirror
Is Flash Really On 99% of Net Devices?
Barence writes "Adobe claims that its Flash platform reaches '99% of internet viewers,' but a closer look at those statistics suggests it's not exactly all-encompassing. Adobe puts Flash player penetration at 947 million users out of a total 956 million internet-connected devices, but the total number of PCs is based on a forecast made two years ago. What's more, the number of Flash users is based on a questionable internet survey of just 4,600 people — around 0.0005% of the suggested 956,000,000 total. Is it really possible that 99% penetration could have been reached? Including Linux users? Including users at work? Including brand-new systems?"
I'm proud to be in the 1% of people who haven't been penetrated by Flash.
Flash was originally crafted with the best of intentions, I'm sure, but due to gross misuse by virtually everyone who's ever touched it, Flash has become a blight on the face of the Tubes. Whether it's noisy and annoying ads, embedded-but-not-linked video, site navigation without a plain HTML version, or malware-pushing securityless redirects, Flash has earned its rightful place in /dev/null.
I do browse the Internet with my BlackBerry Pearl, and no, I cannot get Flash video on it. Was the definition of "Internet viewer" tailored to purpose, by any chance?
It would be more interesting to conduct a survey to find what percentage of net users find Flash as annoying as I do.
Maybe some download stats for Flashblock? I would have to agree that Flash is now more or less inescapable, especially if you like YouTube, but if a site is built on Flash it's usually a surefire sign that the content won't be worth the bandwidth.
Please, this "how can just 4600 people represent so many" comment is something any college-educated person should know better than to say. Provided the sample was drawn randomly from a representative pool of users, 4600 people is more than adequate, giving a sampling error of about 2%.
The fact that the sample is a very small fraction of the total population does not make it meaningless.
It may be meaningless for OTHER reasons of course...
The survey was made in flash
The Wise adapts himself to the world. The Fool adapts the world to himself. Therefore, all progress depends on the Fool.
Saying that Flash can be viewed by "99% of internet viewers" is not saying that Flash is on "99% of internet devices." My Centro doesn't have Flash, but my work laptop does, so I'd say "yes" if polled about whether I have access to Flash content. My ratio of internet devices to Flash-capable devices (5:2) doesn't interest Abobe.
Their claim is probably about right.
--I'm so big, my sig has its own sig.
-- See?
[I also posted a portion of this on the original site but thought it might also be useful here.]
Being a Technical Evangelist for Adobe I frequently get questioned about our published statistics. My response is that you should always test YOUR user base before you make a decision about building on any technology. And in most cases when companies do their own testing the results are within one percent of our published numbers. This is true for enterprise's, SMBs, media companies, etc. But occasionally I hear about some demographic where the numbers are totally off. For instance, if your user base is still working on green screens then you will find lower Flash Player penetration numbers there.
I think Slashdot should publish their stats about their users. It would be interesting to see what the Flash Player penetration is like with this demographic - especially considering I sometimes see Flash banner ads on Slashdot.
-James (Adobe)
You could just go the HTML route...
w3c claims that its HTML platform reaches '100% of interweb viewers,' but a closer look at those statistics suggests it's not exactly all-encompassing. w3c puts HTML-capable web browser penetration at 956 million users out of a total 956 million internet-connected devices, but the total number of PCs is based on a forecast made two years ago. What's more, the number of HTML users is based on a questionable internet survey of just 4,600 people - around 0.0005% of the suggested 956,000,000 total. Is it really possible that 100% penetration could have been reached? Including Linux users? Including users at work? Including brand-new systems?
I'd expect tech readers to have a modicum of statistical sense, but the arguments presented in the summary display an embarrasing ignorance of established statistical techniques. The central limit theorem - one of the first things taught on any stats course - suggests that the sample size is more than adequate, and the researchers have made a serious attempt to take a representative sample across coutries, age groups and genders.
The flaws in the research are more subtle but aren't picked up in the summary. First, beware of any vendor-funded survey - you can guarantee that the although the underlying facts are probably accurate, the interpretation will spun to the point of incredulity. Also, there's probably good reason to believe that people who take part in email surveys aren't representative of the wider population.
But the real problem is that the survey muddles up devices and people: the research discovered that 99% of people can read see Flash animations, but that doesn't remotely mean that 99% of internet-connected devices have Flash. My phone is connected to the innternet, but it certainly can't read Flash files, for example, but I generally read emails on my PC not my phone
Having said that, the results smell about right. Almost all PCs have Flash because it's so easy to install these days - even on Linux./P
plus the survey was conducted using a flash pop-up
Some drink at the fountain of knowledge. Others just gargle.
How do they survey the people they can't reach or only speak something like Vietnamese?
- Zav - Imagine a Beowulf cluster of insensitive clods...
Googles Analytics does check for flash player. I'm sure they know. ;)
And let's see...956-947 million for a difference of 9 million users? Let's ignore linux users for a moment... we have no idea how many linux users have flash. OTOH, I'm pretty sure I read that Apple has sold well over 10 million iPhones. We all know iPhones don't have flash. So I'm pretty sure we already know that number is absolute bullshit.
Nice work Adobe.
The use of two year-old data for projecting the current Internet population may or may not be questionable, but there's nothing at all wrong with extrapolating from sample of 4600 to a population of one billion -- or any size.
It's a curious result of probability theory that, assuming your sample is truly random (which is HARD to achieve!), the sample size you need is independent of the size of the population you're examining. It doesn't matter whether there are a million, hundred million or hundred trillion Internet-connected computers, a random sample of 4600 is equally good.
Yes, this is counterintuitive, like so much else in probability theory.
When choosing an appropriate sample size what matters is the rarity of the trait you're searching for, the margin of error you want to allow, and the degree of confidence you want to have in your result. It's an interesting circularity that you need to know how common computers without Flash are in order to determine how large a sample you need to determine how common computers without Flash are. In practice it isn't a big deal, though. You guess at your answer, compute the required sample size, perform your sampling operation, then see what answer the sample provided. If it's not close to your assumed answer, then you use the sample as the basis of a new assumption and compute a required sample size for your desired level of confidence. If needed, you sample some more. Usually, though, you can make a good enough initial guess that one round is sufficient.
This is why pollsters can give 3% error margins and 95% confidence intervals for voter preferences even though there are many millions of voters and they only ask a thousand or so. The fact that getting good random samples is so hard explains why pollsters nevertheless do get it wrong from time to time. But asking more people wouldn't help, since the additional samples would likely have the same unknown bias as the first thousand -- or perhaps if they were chosen a different way they'd have a different unknown bias.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
I wish people would understand sample size before claiming it as a weak point.
A sample is by definition a small slice of a population.
Sample size is not a weak or strong aspect of the analysis based on what its ratio is to the entire population, it is weak or strong based on the confidence level you end up with.
Example: If you have a sample size of 4300 people and your resultant confidence level is 99%, then even if you have a total population of 6 billion people, your sample size is perfectly fine. If, however, your confidence level is 95%, the statistical results should be questioned as to their accuracy wrt the entire population.
I don't know what the confidence level of Adobe's survey was, but the summary should not be throwing its results into question based on the sample size to total population ratio, it makes the person sound stupid.
Flash is pretty ubiquitous. It comes on every Windows PC loaded with Internet Explorer, and it's an easy download for Macs and Linux machines. The Android phone OS from Google supports Flash, and Adobe has announced a working Flash for iPhone, simply awaiting Apple's go-ahead. The new Palm Pre phone will have Flash. Windows Mobile has Flash Lite. Probably, Apple will allow Flash if Pre and Android phone sales take off.
Youtube pretty much ensured Flash's predominance. Suddenly, there was an easy, painless way to watch video and listen to audio without having to mess with Realplayer and Windows Multimedia codecs and stupid digital rights management code that only works in certain versions of MS Windows.
It's interesting how Flash took the web app market away from Java. Flash is the big player in interactive web apps, while Java is a bit player. Java is still huge in server side apps but it's dead on the desktop. I can't even get Java applets to run on my current home machine with Firefox and Suse Linux, but I have no incentive to get it working. There are still a couple of web sites out there that use java applets for their user interface widgets, but these are few and far between.
it's = "it is"; its = possessive. E.g., it's flapping its wings.