I was going make the Subject line: Mixing Apples with Oranges, Cheese with Chalk, and Newbies with Virgins
...but realized that, in the Forum Index LIST, it would look meaningless. After reading this you MIGHT understand why I wanted that Subject line.

.
Does anyone else agree/disagree that websites - like Distrowatch (which I like, A LOT) - which have a "LISTING" of distros, that is RANKED (ie ordered in some way), should show exactly which version of the distros they are ranking?

For instance if I was making up the list I wouldn't rank (say) Mint-18 alongside Korora-24. Why? Well on the list I cannot tell if the Mint is 32-bit or 64bit. I cannot even tell if it is Mint 14,15,16,17 or 18 that I am looking at. It's just "Mint".

So is it just a total "popularity download" ranking that is used? That is, is everything lumped together? Is it moderated by currency (ie current) or versions or what. I cannot tell. Anyway, if there are "lumpy bits" then the list is skewed for information.

Here's a couple of further aspects trying to demonstrate "why" this IS a problem...


JUST RELEASED, ie NO RANKING...

Consider Fedora-24 which has just recently been released. Is the list showing only Fedora-24, or still Fedora-23, or Fed-23 PLUS Fed-24?

VERSIONS...
Is Fedora-24 (whatever that is) compared to (say) Mageia? Which Fedora; which Mageia? Mageia-5, or Mageia-6 beta, or Mageia-6 the stable, non-release version?

"ODDITIES", THAT ARE NOT MAINSTREAM...
PuppyTahr-6 is only i386 32-bit (perhaps good enough) and is a noteworthy runner in the pack, but other Puppy versions are based on different distros. ALL still "Puppy"?
[NO offence meant to Puppy in that. I use it in certain circumstances, and it is pretty "complete".]

CHUNKYS...
Nor would I compare Mageia-5 i586 32-bit with Mageia-5 64bit. It's not the same thing. [Now Mageia might cast a glaring eye at me over that, and I would have to stand to be corrected, but I meant no harm.]

TRIES
This is a tough one, because it refers to "continued solid usage" (perhaps "usability") - but also INFERS "just trying it out, then dumping it in the rubbish bin".
The stories to demonstrate this alternate "feature" of some distros abound, so I won't continue with it.

.
GENERALIZED INFO
Here is just a rough sampling of distros so you get the drift of my questioning...

Mageia-5 has i586 and 64bit
Debian-8 is based on amd64, arm64, armel, armhf, i386, mips, mipsel, powerpc, ppc64el, s390x
Ubuntu-14.04 (bloody Ubuntu, Ubuntu everywhere) has amd64, i686, powerpc, ppc64el
PCLinuxOS-2016.03 (on Distrowatch) shows i586, although PCLinuxOS says it has dropped 32-bit support. So which PCLinuxOS is it??
OpenSuse now has v42.1 which is 64bit only, but the common v13.2 is i586 and 64bit, yet Tumbleweed (rolling?) is armhf, i586 and 64bit.
Mint-18 is i386 and 64bit.
Fedora-24 has i686, 64bit, armhpf.
Manjaro-16.06.1 is i686 and 64bit.
Antergos-16.06.18 has an i686 and 64bit.
Antix-16 is an i486 (but non-systemd, but can also be systemd configurable).
Black Lab-7.6.1 i386 and 64bit.
Gentoo-20160514 x86, alpha, amd64, arm, hppa, ia64, mips, ppc, ppc64, sparc.
Bohdi-3.2.0 is i386 and 64bit.
TinyCore-7.1 is i486 and 64bit.


NOTES
* This is definitely NOT a flame thing. I'm asking you about how this stuff gets viewed and used, and what is actually compared.
* I'm not even talking about "DistroWatch" because I like what they are trying to do (ie doing). I am refering to ANY distro ranking, ANYWHERE.
* I'm definitely not dumping on any distro.
* I have definitely not counted in DMs or DEs in this discussion because THAT might create an argument.

.
I suggest that...

Perhaps the ranking needs to be a kind of Marquis of Queensbury "thing" - and with "divisions". Heavyweights with heavyweights; welterweights with welterweights; skinny kids with skinny kids.

Yes; it would be "a challenge". Did a challenge ever stop anyone in Linux before?


What's Your opinion on: "DISTROS BY RANKED DIVISIONS"?

Regards,
aka boombaby

Since you mention DistroWatch but no other site.

https://distrowatch.com/dwres.php?resource=popularity

It would be quite a challenge and you would have to include DE/WMs. linux is just the kernel. The distribution determines the rest i.e. user interface, installation process and application support...

I'd say that the popularity ranking is such an uninformative thing that the sort of precision you envisage is hardly worth the effort, even if the data could be collected.

Does popularity really matter when one choses a distro? The two I use are listed at 9 and 82, but I wouldn't use any of the top 5.

Although it wouldn't be a factor for some folk, and despite the two not necessarily going hand in hand, the fact that more popular distros tend to have a greater quantity of online support (forums etc.) would be a factor for me personally, as a relative newbie, when choosing a distro.

Hello, ....

michaelk said...

Firstly, I love DistroWatch, and there is no "evil" intent in any of my comments. I think Jessie is great, and I enjoy reading his comments. I lurv DistroWatch!

Also quoted website...
Secondly, yes, I knew (roughly) about that comment, but seeing the distros listed "1 to n" down the Home page gives a certain kind of "impression". Seeing an empirical value for downloads there gives a certain kind of "impression". You, yourself, must get some kind of "impression" from that list. Indeed, if (in that list) I hold my mouse over the little pivot on the right (ie the up/down/none indicator) you will see that yesterday there was "x" downloads, which gives a further solid impression. My point was/is for say Fedora-24 released recently what does that Fedora number mean? Fedora-24, or Fedora-23+24, or... ...well I won't go over it.

Also said...

To set up a system to count a 64bit download separate to a 32bit download, or the last distro separate from the latest distro isn't like "Men-In-Black 2" where a bug-eyed critter sits under the desk counting them out with beans as they get downloaded. For crying out LOUD, WE'VE GOT BUG-EYED LINUX TO DO THE COUNTING! That's what computers do best! Well, sort of best. (I guess the only question there is, "Which version would be best to do the counting?".)

DavidMcCann said:
Sorry, David(McCann), you can't say the distros you use are "listed at 9 and 82"! Which version do you have? Good! And WHICH version is listed at 9? Which at 82? You cannot tell. See?

.
Like Member "hydrurga", for me popularity is pretty damn important - and I will demonstrate with 3 examples...

(1.) I recently stopped using a truly great distro - PCLinuxOS (32-bit) - because they dropped unpopular 32-bit support, for the popular 64bit. I couldn't download the Office pkg and install it, so finito.

(2.) I recently had to stop using Win2000 because websites were no longer supporting the older, unpopular browser that I had to use. Nor were developers updating old, unpopular software for it. People had chosen (with MS pushing) to move to popular XP > Vista > 7 > 8 > ? (For me Win2K was the best.)

(3.) I'll quickly tell you this story too. I once studied PC's at tech school. I had a 32-bit Atari-ST with 1MB of RAM, a 3-1/2" disk drive. It also had a full 32-bit bus, and it flew! The graphics were very hi-qual. At that time it beat ANYTHING else around for technical ability. I also managed to pick it up 2nd-hand with several pretty decent games too. One day I was extolling the virtues of such a system to my much-younger classmate. He had an IBM clone with 64kb of RAM and a 5-1/4" floppy drive. It had an 8-bit bus, and (from memory) CGA graphics. [I might be wrong; it might have had VGA graphics (but I don't think so.] We had some arguments back and forth about technical aspects when he said, "Mine is much better than yours for ONE reason."

"What?", I asked.

He said, "Mine is more popular, so it will just keep going on and on with hardware and software. Yours isn't very popular, so it will die off one day!" Was he right? (Little B...!)

.
If I talk of popularity I'm NOT talking about a popularity "contest"; I'm talking "numbers". Just "number-crunching". Then, I can seek out "why?" that one is being chosen more than the other one. There has to be a reason. At the moment that "reasoning" isn't being demonstrated clearly, and I believe that leads to an incoherent choice (by Users/Public), and that is one reason (ie ONE reason) why Linux distros now seem to be higgledy-piggledy in quality from one release to the next, one distro to the next, one year to the next. (That is to say, sometimes, but not always.)

I reackon the ones near the top are there mostly because they are based on a distro - DEBIAN - which is generally recognized by people as "stable". Even with the version "hidden", the selection will be vaguely coherent. Further down the list, it's a very hard fight to the top. [For instance, PCLinuxOS is probably demonstrating its nice qualities therein as it works up the chart, I'd say. However, I've only used the excellent 32bit, not the 64bit version. But which is which on the distro chart anyway?]

.
I'll stop now, and hand it back to YOU. Hey, YOU! Out there! Yes, You! Got a comment?

Regards,
aka boombaby

The only way to get accurate stats on which precise distros+versions+flavours are in use would be to have all Linux machines phone home with this info to a central server at some point. Or carry out some form of targetted large-scale statistical sampling on Linux box operators.

In other words, I don't think so.

1 members found this post helpful.

The distrowatch ranking has nothing to do with versions, usage or downloads. It is only the count of how many people access that particular page on its website a day. On that particular distribution web page it contains the available architectures. In my opinion with the way distrowatch is organized it would be very confusing to have something based upon architecture.

.
So, michaelk, here's a little exercise to think about...

I'm standing on a corner recording cars going past. I create a list, and publish it on a website. I update the list daily. The list has:- Ford, Jaguar, MiniCooper, Bugati, DeLorean, Daimler-Benz, Hummer, Daewoo, Reliant Robin, Gogomobile, and some others.

Someone asks me, 'What's a good car to buy?'

I say, 'Have a look on "Cars-o-million"; there's a list there. By the way, Ford is pretty popular at the top of the list. Buy a Ford.'

Anything meaningful in that? To me, not really. If you think carefully about it you might come to a similar conclusion.

If you can see something especially meaningful in the "cars" example just given - that is to say as a generic "list", WITHOUT the additional criteria applied - then please explain that too.

Would the list improve, or be more meaningful, if I mention "model", "engine size", "performance", "economy", "features"? "Cost" might be good? These might reflect a bit better on the Ford (or, indeed, on ALL the cars), wouldn't it. Wouldn't that information be better?

[Knit-pickers might now chime in and ask, "What about cab-size, seats, gears, gearing, towbar, chrome fuel-cap, etc., etc? So to help short-circuit that diversion, I would answer, "Yes; include that if you can - but if you can't then including the primary criteria would make a great start/difference for the list."]

I acknowledge that some websites list a lot of distros, and often have oodles of excellent information about them too. What a list on a website does is give additional meaning. My point is about "useful meaning".

michaelk, there is nothing personal in this, and no offence is intended. It is just an exercise in potentially improving stuff we have in Linux.

Regards,