Hi, I agree with Kristian. Given the way it works, the statistics are really meaningless and I feel you shouldn't drive important choices based on bad statistics. I personally would suggest displaying a link to a feedback/survey form with web downloads and display a message after rpm/deb installation that says something like "please visit http://blah/blah/blah/survey to tell us more about the features you use and help direct the future development of MariaDB". This has an added bonus: not all users know about all features, and a list/survey of the important and interesting ones could get more users to use them. Just my $.02 --Justin On Mon, Mar 9, 2015 at 1:19 AM, Kristian Nielsen <knielsen@knielsen-hq.org> wrote:

Michael Widenius <monty@askmonty.org> writes:

...

for the alpha so I suggested Sergei today that we should enable it for the beta period of MariaDB 10.0

(10.*1* beta, I guess?)

...

As most MariaDB users should know, the feedback is totally anonymous and no private or sensitive information is being sent.

Any comments, suggestions or recommendations?

I think it is a bad idea. Please do not do it.

"Phone-home" is a misfeature in any product, and even more so in system software like a database.

And besides, the information is much less useful than you think, because of unknown, but probably extreme, data skew. In fact, it will probably be more harmful than useful because people will use bad data to justify bad decisions.

Experience supports this point of view with our download numbers. They do not include apt-get / yum / etc. installations, which judging from IRC conversations are the majority. Yet people continuely refer to them as though they mean anything, just because they are there.

- Kristian.

_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp

Hi, Adam! On Mar 09, Adam Scott wrote:

Maybe make it an option when installing?

Yes, that'd be great. On Windows there's a GUI installer, and it has a checkbox for feedback plugin. That's why we get >95% of reports from Windows. Most users don't mind having it enabled, so it seems. So if rpm/deb packages would ask about feedback - that'll help a lot. It just needs be done carefully not to break unattended installs. Regards, Sergei

I understand that the statistics would have some meaning, and understand that there's no actual data being sent, etc. However, enabling the feedback plugin by default IS going to create a whole lot of negative publicity for MariaDB. Is the limited data the feedback plugin provides worthwhile? Look at the similar situation with Ubuntu/Canonical. No matter how much they argued there was no real risk, they suffered a huge amount of negative publicity and eventually reversed the decision. You can argue that the data passed in that case was far more invasive and that they're unrelated, but there WILL be people who have a valid reason to be upset that their IP etc. has been unknowingly passed on, and saying that "you should have switched it off" or "you shouldn't have been using a beta version" is not a valid response after the fact. And then there will be even more people who have no reason to be upset, but will be anyway. Why go through this pain? Rather use this as an opportunity to create a thorough survey (which can provide far more detailed and useful information than the plugin running on a beta version) and engage constructively with the community. The survey has other benefits too, as pointed out elsewhere in the thread - raising awareness of lesser-known features, and just spreading the survey will be marketing for MariaDB. And by all means, give people an option to enable the plugin on installation, as long as it defaults to off. ian On 10/03/2015 13:54, Sergei Golubchik wrote:

A survey is a pretty good idea, thanks!

It may not provide a much better (as in "representative sample") statistics, but it will surely tell the users about the features.

Hi, Ian! On Mar 11, Ian Gilfillan wrote:

I understand that the statistics would have some meaning, and understand that there's no actual data being sent, etc.

However, enabling the feedback plugin by default IS going to create a whole lot of negative publicity for MariaDB. Is the limited data the feedback plugin provides worthwhile? Look at the similar situation with Ubuntu/Canonical. No matter how much they argued there was no real risk, they suffered a huge amount of negative publicity and eventually reversed the decision.

Yes. This is my worry too. That's why Monty has sent this email, that's why it's planned to be enabled only for beta. But asking people to "please enable the feedback plugin" just doesn't work. It seems that most of our users don't care about it at all - I mean, they don't care whether it's enabled or not, and don't care to do anything. That is, if the plugin is disabled - it'll stay disabled, that's what is happening on Linux. If it's enabled - it'll stay enabled, that's what is happening on Windows (MariaDB Windows installer has a special dialog about it with a checkbox that's enabled by default, if I'm not mistaken).

Why go through this pain? Rather use this as an opportunity to create a thorough survey (which can provide far more detailed and useful information than the plugin running on a beta version) and engage constructively with the community. The survey has other benefits too, as pointed out elsewhere in the thread - raising awareness of lesser-known features, and just spreading the survey will be marketing for MariaDB.

Yes, survey is a good idea and I agree that we should do it.

And by all means, give people an option to enable the plugin on installation, as long as it defaults to off.

Okay, I'll see how we could pop up a dialog box during installation to give users an option to enable or disable the plugin. With the plugin being automatically disabled in non-interactive installations. Regards, Sergei

Sergei,

But asking people to "please enable the feedback plugin" just doesn't work. It seems that most of our users don't care about it at all - I mean, they don't care whether it's enabled or not, and don't care to do anything. That is, if the plugin is disabled - it'll stay disabled, that's what is happening on Linux. If it's enabled - it'll stay enabled, that's what is happening on Windows (MariaDB Windows installer has a special dialog about it with a checkbox that's enabled by default, if I'm not mistaken).

Hmm, I would seriously wonder whether installations that use the default configuration and don't change/manage it will be the best source of data about customized configuration. (Or anything, really.) Is it really that useful to get a bunch of stats about packages that were installed perhaps only to meet some prerequisite for another package? Regards, Jeremy

Hi Serg, On 03/12/15 09:17, Sergei Golubchik wrote:

MariaDB Windows installer has a special dialog about it with a checkbox that's enabled by default, if I'm not mistaken).

I'm sorry, but you are mistaken :) Checkbox is off by default. Also, at some point in the past I compared the download number of MSIs with the number that shows up in feedback plugin stats. The later number (# of enabled feedback plugin) is much less, by 2 orders of magnitude. People seem to be happy with default installer settings, in general.

Hi, Kristian! On Mar 09, Kristian Nielsen wrote:

...

As most MariaDB users should know, the feedback is totally anonymous and no private or sensitive information is being sent.

Any comments, suggestions or recommendations?

I think it is a bad idea. Please do not do it.

"Phone-home" is a misfeature in any product, and even more so in system software like a database.

Agree. I don't like it myself. Still between that and random development decisions driven by the marketing department, I just might prefer phone-home.

And besides, the information is much less useful than you think, because of unknown, but probably extreme, data skew. In fact, it will probably be more harmful than useful because people will use bad data to justify bad decisions.

Right, but there are three approaches to this. First, try to get more reports, in the hope that it'll be a representative sample. That's what we're discussing this email thread. Second, take the known skew into account when analyzing the data. For example, see the OS stats chart (http://mariadb.org/feedback_plugin/stats/os/) - it doesn't mean that 96.3% of MariaDB installations are on Windows, it means that we have disproportionally more reports from Windows. And third, use numbers where the skew doesn't matter. For example, total number of installations.

Experience supports this point of view with our download numbers. They do not include apt-get / yum / etc. installations, which judging from IRC conversations are the majority. Yet people continuely refer to them as though they mean anything, just because they are there.

Of course they mean something. They show, literally, how many times mariadb was downloaded from downloads.mariadb.org. As such, they show that mariadb was downloaded (from all sources) *at least* that many times. Also, one can *reasonably assume* that the number of downloads from all other sources follows the grows of downloads from mariadb.org (unless the number of "other sources" changes). These numbers mean a lot. They just don't mean the total number of all mariadb installation (from all sources). Which is pretty obvious :) Regards, Sergei