Re: percentage of popcon submitters
On Thu, Jan 15, 2009 at 10:00:04PM +0100, markus schnalke wrote: Hoi, I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. My current guess is between 1/3 and 2/3. What do you think? My guess is 42. -- .''`. Aurelien Jarno | GPG: 1024D/F1BCDB73 : :' : Debian developer | Electrical Engineer `. `' aure...@debian.org | aurel...@aurel32.net `-people.debian.org/~aurel32 | www.aurel32.net -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Russ Allbery wrote: [...] what packages on your servers are missing security patches, basically popularity-contest doesn't submit package versions, so it is not *that* easy to know whether security updates have been installed or not. As for what security matters popularity-contest could: * randomly change the recent value of a random number of packages * submit via https (or ftp+ssl), and/or even encrypt the data with gpg * have some sort of apt-pinning so that it is possible to indicate that the data corresponding to a given package(s) or repository (ies) should NOT be sent. thereby preventing the I know when you went on VAC because your xfoo-bar-custom package is marked as old information leak. With those security meassures I believe there's a slight chance that a few more people (or institutions) will install popularity-contest. Cheers, Raphael Geissert -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) iEYEARECAAYFAklzySsACgkQYy49rUbZzlo5rQCffJsZ3Ws3iCrj2XlG47syH+R5 bacAn2tDyPob40e7VdoasMOPL/BBQTt/ =tK0A -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Franklin PIAT wrote: [...] * Some system are unproperly configured and can't submit their popcon (missing http proxy ; smtp server is wrong or blocked by their ISP). Especially when people are travelling. Or there's no internet connection when popcon runs and tries to submit via http so it falls back to mail, but mail is local-delivery only. Cheers, Raphael Geissert -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
At Thu, 15 Jan 2009 22:00:04 +0100,
markus schnalke wrote:
[1 text/plain; us-ascii (7bit)]
Hoi,
I know it is not possible to _know_ the real percentage of uses which
submit popcon stats of all users. But I want to ask for guesses,
because more oppinions do likely improve the result.
My current guess is between 1/3 and 2/3.
What do you think?
I used to try to track down that number using apt-listbugs. From
popcon, I know the amount of users who install apt-listbugs use
popcon. apt-listbugs users can be approximated by unique IPs that are
accessing the Debian BTS.
I think I posted that figure a while ago somewhere.
regards,
junichi
--
dan...@{netfort.gr.jp,debian.org}
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Sat,17.Jan.09, 01:05:47, Kjeldgaard Morten wrote: On 16/01/2009, at 18.27, Johannes Wiedersich wrote: Did you think about thousands of computers having 'private ips' with some nat translation and/or local proxie? (I'm thinking of computer labs, companies, etc. not just the odd home user). Essentially all of the computers at our department share the same public IP. Hundreds of machines accessing proxies, and thousands having private IPs. Are these numbers something you know or are you just throwing them around? Otherwise they can of course be accounted for in the total estimate ;-) I know of at least two ISPs in Romania (one of them being the biggest) who will connect you via NAT and I hear this practice is not unheard of in other countries. You won't see more than a few IPs coming from there... Regards, Andrei -- If you can't explain it simply, you don't understand it well enough. (Albert Einstein) signature.asc Description: Digital signature
Re: percentage of popcon submitters
Bernd Eckenfels e...@lina.inka.de writes: In article 87d4enbfqd@mocca.josefsson.org you wrote: It would establish an upper bound of well-administrated debian machines, I think. It is a lower bound, since I guess there are more cases where more than one machine is updated. The case that you download without need or as a duplicate (With multiple IPs) is very low. I've realized it is not a lower bound either, because some download s.d.o packages to temporary chroots and pbuilds etc which should probably not be counted as a debian machine. Still, trying to get numbers on the various statistics we can easily get at may improve estimates and allow us to follow the trends. /Simon -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
[Markus Schnalke] I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. A while back, someone with access to the download logs for security.debian.org tried to estimate the number of machines downloading security fixes for Debian, based on the assumption that no-one is using a mirror for security fixes. I am unable to find those results using Google right now, but would recommend trying to get hold of those numbers to get new lower bound on the number of Debian installations. As for the answer to your question, I have no idea. :) The amount of contributors to popcon went up a lot when the popcon question started to appear in the default installer in Etch, but I have no idea how many actually replied yes to that question. :) Happy hacking, -- Petter Reinholdtsen -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Fri, 16 Jan 2009 08:45:12 +0100 Kjeldgaard Morten m...@bioxray.au.dk wrote: Thanks. Unless you setup some experimental method, any argument should reduce to handwaving or extension of various particular examples.. Surely, it must be possible to get an estimate of the number of downloads of important packages and security updates? I know these downloads also are requested from mirror sites, but at least for the official mirror sites their relative activity must be known? How do you map the number of downloads to the number of users or machines? I have dozens of chroots that I use for multiple reasons. Now, maybe I should use an apt proxy but most of these are cross-building chroots so that doesn't help as the proxy will have amd64 packages and I need arm or armel etc. Then you have the problem of people who maintain local mirrors (often quite short lived ones). It's just more handwaving - unless you want to count every chroot and every local mirror (per architecture) as a separate user. There is no way of accurate counting unless access to the files is restricted to a known number of download methods that all require user intervention to proceed, at which point Debian would not be free. The LinuxCounter method is completely arbitrary - the figures on the site are guesswork and cannot be used in any other calculations. LinuxCounter tries to extrapolate from 180,000 to 29,000,000 without any real basis for such a leap of faith other than we guessed 18 million some time ago and we have x% increase in our counter figures since then, so increase 18 million by x%. I'm not knocking their figures, just reiterating what is on the linux counter website - reliable figures just do not exist and trying to create them usually results in restricting the freedoms that attract users in the first place. http://counter.li.org/estimates.php Linux Counter is no more or less reliable than popcon - both are wild guesses from different perspectives. popcon is an wild underestimate, counter could as easily be over as under. Nobody knows and in a very real sense, nobody could ever know with any accuracy. popcon is what we have, it is an indicator with known deficiencies that always need to be taken into account when using popcon data as a factor in any packaging decision but popcon, overall, is just more handwaving. I find it amusing that we post popcon % figures to two decimal places when the real error margins are completely unknown but it reflects the size of Debian - if popcon didn't use decimal places, a vast number of packages in widespread usage would have a popcon % of zero. I could say that LinuxCounter is out by 30% or 70% or 150% and there would be no reason to consider my guesses as more or less reliable than the ones from LinuxCounter. The whole thing is a complete unknown. -- Neil Williams = http://www.data-freedom.org/ http://www.nosoftwarepatents.com/ http://www.linux.codehelp.co.uk/ pgpWDE8eJPepc.pgp Description: PGP signature
Re: percentage of popcon submitters
[2009-01-16 10:09] Neil Williams codeh...@debian.org The whole thing is a complete unknown. Of course you're right. But it's the best we have. Instead of leaving it with ``we simply don't know'', I prefer to estimate on the (unsure) data sources that are available. For my case, I received valuable comments and approaches that improved my estimations, so I thank everyone who contributed! FYI: I now assume one third of all Debian users submit their stats, with the remark that one third is probably a high guess as it means there would be only about 230 thousand Debian installations in total. But according to counter.li.org between 490 thousand and 12 million Debian users can be estimated. I later reduce the resulting number by one third to respect users with multiple installations. I think my final result is not too large, which is to avoid. If it would be only 1/10 of the true number, I have no problem with it. thanks again meillo signature.asc Description: Digital signature
Re: percentage of popcon submitters
Le Friday 16 January 2009 11:51:50 markus schnalke, vous avez écrit : [2009-01-16 10:09] Neil Williams codeh...@debian.org The whole thing is a complete unknown. Of course you're right. But it's the best we have. Instead of leaving it with ``we simply don't know'', I prefer to estimate on the (unsure) data sources that are available. Not that I like to be polemic, but this sentence doesn't mean anything. If the answer is we don't know, then we don't know. Problem is that you don't give any ground to your claims, hence it is far worse to give any estimation. The only serious analysis was the one made by Bernd Eckenfels, which ended with 1%. I don't really believe this can be used as it is before another contradictory analysis can be done. Of course, you surely do not need such serious considerations for your precise issue, but claiming that any ungrounded estimation is better than nothing made me tilt :) Romain -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
[2009-01-16 12:06] Romain Beauxis to...@rastageeks.org Le Friday 16 January 2009 11:51:50 markus schnalke, vous avez écrit : [2009-01-16 10:09] Neil Williams codeh...@debian.org The whole thing is a complete unknown. Of course you're right. But it's the best we have. Instead of leaving it with ``we simply don't know'', I prefer to estimate on the (unsure) data sources that are available. Not that I like to be polemic, but this sentence doesn't mean anything. If the answer is we don't know, then we don't know. Problem is that you don't give any ground to your claims, hence it is far worse to give any estimation. IMO that's not right if you want to act. Acting without any plan is worse than acting with an estimation (that bases on the best information available). Imaging navigation within permanent fog. I think it is better to go towards the most probable direction, instead of doing nothing because you dont know. Of course this again is dependent on external factors like dangerous ground ... thus no answer fits every case. In the end it's a philosophical question anyway. ;-) Of course, you surely do not need such serious considerations for your precise issue correct ;-) meillo signature.asc Description: Digital signature
Re: percentage of popcon submitters
Neil Williams codeh...@debian.org writes: On Fri, 16 Jan 2009 08:45:12 +0100 Kjeldgaard Morten m...@bioxray.au.dk wrote: Thanks. Unless you setup some experimental method, any argument should reduce to handwaving or extension of various particular examples.. Surely, it must be possible to get an estimate of the number of downloads of important packages and security updates? I know these downloads also are requested from mirror sites, but at least for the official mirror sites their relative activity must be known? How do you map the number of downloads to the number of users or machines? It would establish an upper bound of well-administrated debian machines, I think. I have dozens of chroots that I use for multiple reasons. Good point. I wonder how much these contribute to the overall statistics though. Alternatively, one could argue relatively convincing that a chroot with a complete debian system should be counted as another debian installation. Compare with virtual machines, which is rather similar to a chroot installation on a normal PC. /Simon -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Hi Noah Slater wrote: On Thu, Jan 15, 2009 at 10:00:04PM +0100, markus schnalke wrote: I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. [..] is like this famous old problem: Nobody was permitted to see the Emperor of China, and the question was, What is the length of the Emperor of China's nose? [..] I am writting a paper on popcon (ranking package usage), which isn't finished yet. However I have already identified the following bias: popcon bias --- **popularity-contest isn't part of any task (but prompted by D-I, since Etch). Therefore it is unlikely to be active on system installed using debootstrap (xen-create-image, etc..) **Votes are purges after 23days of inactivity. * Popcon statistics combines Old-Stable, Stable, Testing and Unstable... and maybe some Debian-derivatives. user/sysadmin bias -- * Installing popularity-contest is optional. * Some system are unproperly configured and can't submit their popcon (missing http proxy ; smtp server is wrong or blocked by their ISP). Especially when people are travelling. * People deploying Debian in large companies usually disable popcon. * Security concerns (Do you think that people who install security audit tools, like nessus, will submit popcon?). * ISPs and hosting companies disables popcon by default. Franklin -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
James Vega james...@debian.org writes: On Thu, Jan 15, 2009 at 4:55 PM, markus schnalke mei...@marmaro.de wrote: [2009-01-15 22:37] Michael Goetze mgoe...@mgoetze.net before wild speculations ensues, you might want to specify what you really want to know: the percentage of people installing debian systems who use popcon (always/sometimes), or the percentage of installed machines that submit popcon data? Seems my wording was unclear. I want to know the percentage of installed machines that submit popcon data. That requires knowing the number of computers that have Debian installed which, as has been discussed various times in the past on this list, is difficult to determine. How about numbers for security.debian.org downloads? That will measure the number of well-administrated debian machines (except those well-administrated machines that use other mirrors). /Simon -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Fri, 16 Jan 2009 13:24:58 +0100 Simon Josefsson si...@josefsson.org wrote: Neil Williams codeh...@debian.org writes: Surely, it must be possible to get an estimate of the number of downloads of important packages and security updates? I know these downloads also are requested from mirror sites, but at least for the official mirror sites their relative activity must be known? How do you map the number of downloads to the number of users or machines? It would establish an upper bound of well-administrated debian machines, I think. No, merely the number of installations which is not the same, clearly. Chroots can be entirely temporary. I regularly hammer the mirrors to create test chroots last a matter of minutes. (Usually in a different architecture each time, hence a proxy isn't much help.) It's not just chroots either - don't forget issues of local mirrors. Download measurements cannot take account of whether the downloaded file is actually installed or merely copied into another repository. I have dozens of chroots that I use for multiple reasons. Good point. I wonder how much these contribute to the overall statistics though. Currently, these are all ignored for popcon but would register in any download measurements - repeatedly. Alternatively, one could argue relatively convincing that a chroot with a complete debian system should be counted as another debian installation. Compare with virtual machines, which is rather similar to a chroot installation on a normal PC. In that case, I'm probably responsible to thousands of 'installations' and those DD's involved in D-I must have an inconceivable number of installs. Frans? any idea how many installs you've clocked up by that measure? Out of all those thousands of chroots and dozens of local mirrors that I've created (and subsequently removed) just since Etch, I've only actually got 6 real machines that use Debian. Six machines that actually boot, six that have real users and need maintenance - only four that are regularly powered on. IMHO, a chroot is not an installation - even if the chroot contains the whole of GPE or the whole of GNOME, it's a chroot. It doesn't have real users, it doesn't boot, it is a test environment only. If you want to count well maintained Debian machines you have to exclude all chroots and all local mirrors. How is it worth recording data from a debug install that lasts only a few seconds after completing the install and which is instantly replaced by yet another test? How is it worth recording any data from downloads that merely result in yet another copy of the original mirror. -- Neil Williams = http://www.data-freedom.org/ http://www.linux.codehelp.co.uk/ http://e-mail.is-not-s.ms/ pgpsi5k7Iz5PB.pgp Description: PGP signature
Re: percentage of popcon submitters
On Fri, 16 Jan 2009 13:21:29 + Neil Williams codeh...@debian.org wrote: In that case, I'm probably responsible to thousands of 'installations' OK, that's an exaggeration but it's certainly hundreds since Etch. -- Neil Williams = http://www.data-freedom.org/ http://www.linux.codehelp.co.uk/ http://e-mail.is-not-s.ms/ pgpSyNMvRhBL0.pgp Description: PGP signature
Re: percentage of popcon submitters
El Vie 16 Ene 2009, Simon Josefsson escribió: How about numbers for security.debian.org downloads? That will measure the number of well-administrated debian machines (except those well-administrated machines that use other mirrors). well-administrated *etch* machines. luciano -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
* Luciano Bello [Fri, 16 Jan 2009 11:37:39 -0200]: El Vie 16 Ene 2009, Simon Josefsson escribió: How about numbers for security.debian.org downloads? That will measure the number of well-administrated debian machines (except those well-administrated machines that use other mirrors). well-administrated *etch* machines. Hey, Lenny too. -- Adeodato Simó dato at net.com.org.es Debian Developer adeodato at debian.org There is no man so good who, were he to submit all his thoughts to the laws, would not deserve hanging ten times in his life. -- Michel de Montaigne -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Neil Williams codeh...@debian.org writes: On Fri, 16 Jan 2009 13:24:58 +0100 Simon Josefsson si...@josefsson.org wrote: Neil Williams codeh...@debian.org writes: Surely, it must be possible to get an estimate of the number of downloads of important packages and security updates? I know these downloads also are requested from mirror sites, but at least for the official mirror sites their relative activity must be known? How do you map the number of downloads to the number of users or machines? It would establish an upper bound of well-administrated debian machines, I think. No, merely the number of installations which is not the same, clearly. Chroots can be entirely temporary. I regularly hammer the mirrors to create test chroots last a matter of minutes. (Usually in a different architecture each time, hence a proxy isn't much help.) It's not just chroots either - don't forget issues of local mirrors. Download measurements cannot take account of whether the downloaded file is actually installed or merely copied into another repository. It would still provides an upper bound, but the local mirror exception is a good point. So the number derived from security.debian.org statistics would be 'an upper bound of the number of well-administrated debian installation that do not use local security mirrors'. I assume this number is correlated to the number of real debian installations (although I'm not sure we have a good definition of real here?). Merely the number of distinct IP addresses downloading a particular popular update from security.debian.org at least once would be interesting. /Simon -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Friday 16 January 2009 15:42:38 Neil Williams wrote: On Fri, 16 Jan 2009 13:21:29 + Neil Williams codeh...@debian.org wrote: In that case, I'm probably responsible to thousands of 'installations' OK, that's an exaggeration but it's certainly hundreds since Etch. This is true, but I imagine that this could be alleviated by taking into account some popular file modification time, like /etc/debian_version? If found to be young enough, then we can count that as a throw-awayer we don't care about, otherwise it is a valid entry, no matter how and for what reasons the installation has been created. Still, I don't see a good way to measure non-internetworked users (imagine here evil authorities, secret labs, mysterious scientists performing evil calculations, etc;-) and those who just answered no to popcon's question. We can easily ignore the former (since they are in the 'No Such Entity' subset anyway;-), but the latter matters, and are still adding uncertainty to the equation. It is most likely that the more we beat that, the more unknown quantities will be brought to the scene, hence why bother ;-) -- pub 4096R/0E4BD0AB 2003-03-18 people.fccf.net/danchev/key pgp.mit.edu -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Simon Josefsson wrote: Merely the number of distinct IP addresses downloading a particular popular update from security.debian.org at least once would be interesting. Did you think about thousands of computers having 'private ips' with some nat translation and/or local proxie? (I'm thinking of computer labs, companies, etc. not just the odd home user). Essentially all of the computers at our department share the same public IP. Johannes -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAklww5QACgkQC1NzPRl9qEXRMACeIZ6rTadGokr+Zk5GtLfeglRa GW8AnRij/EmKaY+9o31Hs/TWd6RkL1Lu =EujH -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Petter Reinholdtsen p...@hungry.com writes: A while back, someone with access to the download logs for security.debian.org tried to estimate the number of machines downloading security fixes for Debian, based on the assumption that no-one is using a mirror for security fixes. I am unable to find those results using Google right now, but would recommend trying to get hold of those numbers to get new lower bound on the number of Debian installations. It's worth bearing in mind that that's a bad assumption, too. We use a local security mirror in full knowledge that it's not recommended, but we watch it closely and will manually sync if need be. We do this because we have systems on IP addresses that are not routable to the Internet and need an on-campus source for all package updates. Having a local mirror is easier for our purposes than using a proxy. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On 16/01/2009, at 11.09, Neil Williams wrote: How do you map the number of downloads to the number of users or machines? I have dozens of chroots that I use for multiple reasons. Now, maybe I should use an apt proxy but most of these are cross-building chroots so that doesn't help as the proxy will have amd64 packages and I need arm or armel etc. Chroots have that effect, but the number of users using chroots is negligible compared to the total number of users. They would artificially boost the number of users, whereas other effects -- for example sites having a local repo -- would tend to lower it. Besides, many users of chroots use apt-cacher like me, which would not add to the count at all. I agree that it's not possible to get an accurate number, but the OP specifically wants an estimate. An estimate of the number of downloads would make it possible to estimate of the fraction of sites that install popcon so a corrective factor could be applied to the popcon statistics. Cheers, Morten -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Fri, Jan 16, 2009 at 11:18:14AM -0800, Russ Allbery wrote:
It's worth bearing in mind that that's a bad assumption, too. We
use a local security mirror in full knowledge that it's not
recommended, but we watch it closely and will manually sync if
need be. We do this because we have systems on IP addresses that
are not routable to the Internet and need an on-campus source for
all package updates. Having a local mirror is easier for our
purposes than using a proxy.
Same here, though with a caching Debian package proxy instead of an
actual mirror. Nonetheless, s.d.o only sees one download of a given
security update even though it's actually being retrieved by
hundreds of machines.
For that matter, we also have a separate network for facility
security systems which is not connected to the Internet, and once a
month we rsync a detachable drive on an Internet-connected machine,
then sneaker-net it back to the system acting as a security update
repository for the other hosts.
--
{ IRL(Jeremy_Stanley); PGP(9E8DFF2E4F5995F8FEADDC5829ABF7441FB84657);
SMTP(fu...@yuggoth.org); IRC(fu...@irc.yuggoth.org#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER(fu...@yuggoth.org);
MUD(fu...@katarsis.mudpy.org:6669); WWW(http://fungi.yuggoth.org/); }
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Johannes Wiedersich johan...@physik.blm.tu-muenchen.de writes: Simon Josefsson wrote: Merely the number of distinct IP addresses downloading a particular popular update from security.debian.org at least once would be interesting. Did you think about thousands of computers having 'private ips' with some nat translation and/or local proxie? (I'm thinking of computer labs, companies, etc. not just the odd home user). Essentially all of the computers at our department share the same public IP. Right, there are many reasons why such a number wouldn't be perfect, but I still believe it would be interesting to know. Especially if you plot the trend in a graph to watch yearly changes. If you get 10 such indicator variables that likely are somehow correlated to the number of machines (virtual or not) running debian plotted in a graph, and watch the trends, that is likely to be the best measure we are likely to ever get. Or are there any better ideas on how to get a good estimate? /Simon -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On 16/01/2009, at 23.25, The Fungi wrote: Same here, though with a caching Debian package proxy instead of an actual mirror. Nonetheless, s.d.o only sees one download of a given security update even though it's actually being retrieved by hundreds of machines. On 16/01/2009, at 18.27, Johannes Wiedersich wrote: Did you think about thousands of computers having 'private ips' with some nat translation and/or local proxie? (I'm thinking of computer labs, companies, etc. not just the odd home user). Essentially all of the computers at our department share the same public IP. Hundreds of machines accessing proxies, and thousands having private IPs. Are these numbers something you know or are you just throwing them around? Otherwise they can of course be accounted for in the total estimate ;-) Cheers, Morten -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
In article 200901161206.13302.to...@rastageeks.org you wrote: If the answer is we don't know, then we don't know. Problem is that you don't give any ground to your claims, hence it is far worse to give any estimation. But if you say we see security donloads from x unique IPs for every new update then you have a lower bound. The only serious analysis was the one made by Bernd Eckenfels, which ended with 1%. I don't really believe this can be used as it is before another contradictory analysis can be done. Well, its not too serious, since Linuxcounter also only estimates the 29 Million Users. Gruss Bernd -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
In article 87d4enbfqd@mocca.josefsson.org you wrote: It would establish an upper bound of well-administrated debian machines, I think. It is a lower bound, since I guess there are more cases where more than one machine is updated. The case that you download without need or as a duplicate (With multiple IPs) is very low. Gruss Bernd -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Sat, Jan 17, 2009 at 01:05:47AM +0100, Kjeldgaard Morten wrote:
Hundreds of machines accessing proxies, and thousands having
private IPs. Are these numbers something you know or are you just
throwing them around? Otherwise they can of course be accounted
for in the total estimate ;-)
I could count mine with wc -l (I have to maintain a list to kick off
aptitude runs on them all). As for Russ, I'm sure he could at least
analyze his mirror's access logs for a more precise value if he
wanted to, just as someone with access to the s.d.o logs might. The
bigger issue is that there are almost certainly enough sites like
ours to make them a drop in the bucket, skewing the measure downward
fairly significantly.
Economies of scale, security, bandwidth costs, IPv4 runout, et
cetera assure that the more Debian machines running at a given
location, the less likely it is that you'll be able to count them
externally without popcon or somethig similar installed (and as has
been discussed, these are the same ones who are less likely to
install it site-wide anyway). The most I expect you can hope for
here is a relatively reasonable lower bound on actively-updated
systems. If there are probably at least X number of Debian systems
on the Internet, but maybe many more isn't a sufficiently useful
statement for you, you might be out of luck.
--
{ IRL(Jeremy_Stanley); PGP(9E8DFF2E4F5995F8FEADDC5829ABF7441FB84657);
SMTP(fu...@yuggoth.org); IRC(fu...@irc.yuggoth.org#ccl); ICQ(114362511);
AIM(dreadazathoth); YAHOO(crawlingchaoslabs); FINGER(fu...@yuggoth.org);
MUD(fu...@katarsis.mudpy.org:6669); WWW(http://fungi.yuggoth.org/); }
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Hi, I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. My current guess is between 1/3 and 2/3. What do you think? before wild speculations ensues, you might want to specify what you really want to know: the percentage of people installing debian systems who use popcon (always/sometimes), or the percentage of installed machines that submit popcon data? For example, I'm pretty sure any hosting company offering Debian on dedicated servers will disable popcon by default... Regards, Michael -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
[2009-01-15 22:37] Michael Goetze mgoe...@mgoetze.net before wild speculations ensues, you might want to specify what you really want to know: the percentage of people installing debian systems who use popcon (always/sometimes), or the percentage of installed machines that submit popcon data? Seems my wording was unclear. I want to know the percentage of installed machines that submit popcon data. Maybe some words about the background of this question: I want to estimate the number of users of some software. Thus I have a look at Popcon which tells me the number of installations of the package. Now I need a good multiplicator (the searched one) to receive an estimated number of users within Debian. I'll do the same for Ubuntu's popcon and add guessed numbers for usage on other GNU and Unix systems. That should lead to a quite good estimation. For example, I'm pretty sure any hosting company offering Debian on dedicated servers will disable popcon by default... And there is a number of systems without online connection, they will also not submit. meillo signature.asc Description: Digital signature
Re: percentage of popcon submitters
On Thu, Jan 15, 2009 at 4:55 PM, markus schnalke mei...@marmaro.de wrote: [2009-01-15 22:37] Michael Goetze mgoe...@mgoetze.net before wild speculations ensues, you might want to specify what you really want to know: the percentage of people installing debian systems who use popcon (always/sometimes), or the percentage of installed machines that submit popcon data? Seems my wording was unclear. I want to know the percentage of installed machines that submit popcon data. That requires knowing the number of computers that have Debian installed which, as has been discussed various times in the past on this list, is difficult to determine. -- James GPG Key: 1024D/61326D40 2003-09-02 James Vega james...@debian.org -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
James Vega wrote: On Thu, Jan 15, 2009 at 4:55 PM, markus schnalke mei...@marmaro.de wrote: I want to know the percentage of installed machines that submit popcon data. That requires knowing the number of computers that have Debian installed which, as has been discussed various times in the past on this list, is difficult to determine. And even then it might not help answer your question, for instance if you have a desktop application the percentage of popcon submitters might be higher than average among your users, whereas if you have some software mainly useful on classified military machines, the percentage of popcon submitters might be lower than average among your users. -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Michael Goetze schrieb: Hi, I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. My current guess is between 1/3 and 2/3. What do you think? before wild speculations ensues, you might want to specify what you really want to know: the percentage of people installing debian systems who use popcon (always/sometimes), or the percentage of installed machines that submit popcon data? For example, I'm pretty sure any hosting company offering Debian on dedicated servers will disable popcon by default... Hello, for myself I deactivated popcon on every machine. Then I first installed and activated it on every server and some times later also on my desktop. For my case I do not think that it will be realy deactivated on dedicated servers in most cases. But yeah, it may differ from person to person :) -- /* Mit freundlichem Gruß / With kind regards, Patrick Matthäi E-Mail: patrick.matth...@web.de Comment: Always if we think we are right, we were maybe wrong. */ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
El Jue 15 Ene 2009, markus schnalke escribió: My current guess is between 1/3 and 2/3. that means that there is between 78055/(1/3)=234165 and 78055/(2/3)=117,082 of Debian installations. It doesn't look like a big number... I think that we are more. Maybe your estimation is too high. luciano -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
In article 20090115210004.gv21...@serveme.schnalke.local you wrote: My current guess is between 1/3 and 2/3. Machines or Users? According to Linuxcounter there are estimated 29,000,000 users and debian has 18.36% which equals in 5m debian users. Popcon lists 78k submissions, which is less than 2% Gruss Bernd -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
markus schnalke mei...@marmaro.de writes: I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. My current guess is between 1/3 and 2/3. What do you think? We run Debian or Ubuntu on around 500 machines and only submit popcon results from about four of them. It's hard to talk institutional computer security departments into the idea that the minor risk of the information released by popcon (what packages on your servers are missing security patches, basically) is worth it for just the warm fuzzy feelings of contributing. I suspect that popcon is not running on most large institutional Debian installations. -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
Le Thursday 15 January 2009 23:25:02 Bernd Eckenfels, vous avez écrit : In article 20090115210004.gv21...@serveme.schnalke.local you wrote: My current guess is between 1/3 and 2/3. Machines or Users? According to Linuxcounter there are estimated 29,000,000 users and debian has 18.36% which equals in 5m debian users. Popcon lists 78k submissions, which is less than 2% Thanks. Unless you setup some experimental method, any argument should reduce to handwaving or extension of various particular examples.. The big question (and the big troll) that's hidden behind this question is the total amount of installed debian systems. Since this value is always and always discussed, I don't think there is any broadly accepted counting method, hence I don't think the original question can be answered... Or, if you really want to troll, let's switch to the total amount of installed debian systems, since this is equivalent, but the de^C^Cbattle should be more fun... :-) Romain -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
On Thu, Jan 15, 2009 at 10:00:04PM +0100, markus schnalke wrote: I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. This question of trying to figure out whether a book is good or bad by looking at it carefully or by taking the reports of a lot of people who looked at it carelessly is like this famous old problem: Nobody was permitted to see the Emperor of China, and the question was, What is the length of the Emperor of China's nose? To find out, you go all over the country asking people what they think the length of the Emperor of China's nose is, and you average it. And that would be very accurate because you averaged so many people. But it's no way to find anything out; when you have a very wide range of people who contribute without looking carefully at it, you don't improve your knowledge of the situation by averaging. -- Richard P. Feynman, Surely You're Joking, Mr. Feynman! -- Noah Slater, http://tumbolia.org/nslater -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Re: percentage of popcon submitters
[2009-01-16 05:59] Noah Slater nsla...@tumbolia.org On Thu, Jan 15, 2009 at 10:00:04PM +0100, markus schnalke wrote: I know it is not possible to _know_ the real percentage of uses which submit popcon stats of all users. But I want to ask for guesses, because more oppinions do likely improve the result. This question of trying to figure out whether a book is good or bad by looking at it carefully or by taking the reports of a lot of people who looked at it carelessly is like this famous old problem: Nobody was permitted to see the Emperor of China, and the question was, What is the length of the Emperor of China's nose? To find out, you go all over the country asking people what they think the length of the Emperor of China's nose is, and you average it. And that would be very accurate because you averaged so many people. But it's no way to find anything out; when you have a very wide range of people who contribute without looking carefully at it, you don't improve your knowledge of the situation by averaging. -- Richard P. Feynman, Surely You're Joking, Mr. Feynman! Good point, but one may refer to the Delphi method: http://en.wikipedia.org/wiki/Delphi_method However, the answers I received actually helped my. Not because they were estimations, but because they were comments for what to keep in mind. In any way, I believe more oppinions do improve results, but not by telling numbers one can average, but by showing how others see the situation. This widens the own view. meillo signature.asc Description: Digital signature
Re: percentage of popcon submitters
[2009-01-15 23:25] Bernd Eckenfels e...@lina.inka.de In article 20090115210004.gv21...@serveme.schnalke.local you wrote: My current guess is between 1/3 and 2/3. Machines or Users? Popcon focuses on machines. In the end I want users. But any number would be good. According to Linuxcounter there are estimated 29,000,000 users and debian has 18.36% which equals in 5m debian users. Popcon lists 78k submissions, which is less than 2% That is really a good approach. Thanks for that! I seems to be quite sure that the popcon submitters are less than 1/3 of all Debian users. Luciano Bello's calculation pointed to a similar way. meillo signature.asc Description: Digital signature
Re: percentage of popcon submitters
Thanks. Unless you setup some experimental method, any argument should reduce to handwaving or extension of various particular examples.. Surely, it must be possible to get an estimate of the number of downloads of important packages and security updates? I know these downloads also are requested from mirror sites, but at least for the official mirror sites their relative activity must be known? Cheers, Morten -- Morten Kjeldgaard m...@ubuntu.com Ubuntu MOTU Developer GPG Key ID: 404825E7 -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
