3

I wanted to ask a question about software usage statistics here, but other people don’t seem to like my question, for reasons I can’t really understand. Is there anywhere on Stack Exchange where I can ask such a question?

  • 1
    Read up on stats.stackexchange.com and its tour, etc. and go into a chat room there perhaps or find a mod's answer or question and ask them about putting this there. Maybe reading over the tour or help on that SE community will confirm but potentially worth a read at least just in case. – Pimp Juice IT Feb 23 '18 at 20:37
  • 1
    I would say the same thing about User Experience Stack Exchange. I actually doubt that your question would be on-topic there, but they might be receptive to discussing your concerns on Meta. – Scott Feb 23 '18 at 21:54
8

Statistics about software usage are transient. This means the actual data is ever-changing. By contrast, answers on Super User are permanent.

Good answers should be just as valid 10 years down the road as they are right now. This is often true by the scoping of a question. For instance, a question involving compatibility of certain 2008-era Intel hardware (say, a Core 2 processor) and Windows XP is still valid in 2018, 2028, and even a thousand years in the future, because the fact of the compatibility between these parts will never change so long as the hardware continues to run and the bytes comprising the software continue to be available.

Statistics are a snapshot in time of a particular survey, with all the biases (known and unknown) of the collector of the data. They are not a good fit for repositories of knowledge like Super User, because the question can never definitively be answered.

You could scope your question as, "What percentage of Internet users currently use IE as of February 2018?" This would be a "timeless" question of sorts, because the truth about browser usage in February 2018 will not change after February 2018 has come and gone; these become historical facts. For instance, if I told you that 90% of Internet users used Internet Explorer on March 20, 2005, that fact is no less true if you read it in the year 9000.

The problem is, questions about historical facts from a point in time are still pretty bad questions for SU, because they're extremely localized -- meaning, there are unlikely to be many other people who care about that particular question at that particular point in time.

I would argue that by far the most popular point in time for these sorts of statistics is the present. Since "the present" does not refer to any specific point in time, and continues to advance with the passage of time, we can't ask a question whenever the question is being scoped to "the present", because the answers are only going to be true in the instant they are posted. As soon as the answer is posted, and someone else reads it, it's immediately outdated by new data as that data is collected in real time.

And no, browser usage statistics questions are not any more acceptable than calendar usage statistics questions. I don't know where you got the impression that they are.

  • 1
    Thank you. If transiency is a concern for SE question quality, then this gives me more insight into why my question might be considered “bad.” I still think the question would be useful for at least a few years. And answers can always be updated later, as can happen within any other topic, due to the mutating nature of the world. – Jackson Feb 23 '18 at 0:33
  • 1
    As for the “acceptability” of browser usage or calendar usage questions, I never claimed they were “acceptable,” I claimed they are answerable and useful. Apparently adhering to SE sacred law is more important than answerability or usefulness. – Jackson Feb 23 '18 at 0:35
  • 3
    Not everything that can be asked in the form of a question that happens to be useful or interesting to someone can be asked on the Stack Exchange network. That's just the fact of the matter. – allquixotic Feb 23 '18 at 1:35
  • 1
    @Jackson I can answer many questions that are out of scope. Just because I can answer I question doesn’t mean the question should be answered due to it not being a good Q&A question. Some questions are simply not a good fit for StackExchange – Ramhound Feb 23 '18 at 11:27
3

Software usage statistics are almost impossible to give a solid, verifiable and comparable answer for.

There are three types of lies -- lies, damn lies, and statistics.
- Benjamin Disraeli

My ony real question for you it this: how do you go about collecting software usage statistics?

For web browsers this is only achievable because browsers announce themselves to external websites and identify their version in order to allow the website to serve up the most compatible version. That announcement is logged and sent somewhere for comparison but I would argue that the data is skewed. Some people change the version of the browser reported for compatibility reasons, some people change it for privacy or to circumvent blocks based on browser versions.

In fact, in a perfect world, due to your browser being able to be used to track your movements across the web I would argue that it would be beneficial to remove all version information from website requests. This would make broad usage statistics nearly impossible.

You could argue that package download statistics would take over, but a large percentage of people (especially on Linux) avoid downloading directly from source and instead download from their distribution repositories, again skewing the statistics.

So who do we look to for statistics on a broad group of packages that are used in a mostly offline method?

Do we trust the manufacturers of the software to tell us how many users they have? How do they establish how many users they have? Do they potentially invade users privacy and collect "anonymous usage data" (like so many programs these days) or do they use simply count people who bought the product? Do they track how many people download via automatic updates, ignoring those who do not for security reasons such as some high profile recent malware outbreaks?

Once you have data from one manufacturer using one of these methods, how do you compare that to another manufacturer that uses another method? Is it even vaguely comparable?

“Facts are stubborn things, but statistics are pliable.” - Mark Twain

I wouldn't trust manufacturer statistics. They want to make themselves look good. The numbers would be fudged to look as good as possible without being outright made up. Even if you could get statistics from the manufacturers I wouldn't be surprised to find that the total number of users is somehow indicating that every person on the planet uses at least 5 different packages for calendar management.

Some programs lend themselves to easy and impartial usage statistics collection by their method of use, others do not. Neither make for particularly good questions because either the data is there and is trivially accessible, or it is not.

2

nope.

You should only ask practical, answerable questions based on actual problems that you face. Chatty, open-ended questions diminish the usefulness of our site and push other questions off the front page.

Is pretty universal.

  • 2
    In web development, it is very common to refer to statistics on the most commonly-used browsers. e.g. 50% of people use Internet Explorer, so it would be very useful to know that, so one could design websites to work in Internet Explorer. If questions seeking to obtain browser usage statistics are answerable and useful, why wouldn’t questions seeking to obtain calendar software usage statistics also be answerable and useful? – Jackson Feb 22 '18 at 23:47
  • 2
    @Jackson Browser usage statistics are broadly measurable because access requests are logged and correlated from a number of sources. Even so I would question the validity of the data because it is potentially skewed by the nature of the sites where data is collected but at least there is a solid source. Questions based on it are still bad but they are at least workable as data is collectable. Browsers must access external sites. How would anyone collect data on calendar program usage outside of fictitious marketing reports claiming "over 20 billion users" use their product? – Mokubai Feb 23 '18 at 7:11
  • 2
    @Mokubai Other companies could have run into the same problem and polled their users on what they use. It wouldn’t be universally-applicable data but might be representative. – Jackson Feb 23 '18 at 16:21
2

Thanks to the other answerers, I see that the information I seek is probably unobtainable, and for that reason I can see that moderators would not want an unanswerable question cluttering the space of answerable ones.

I think that if I wished to obtain this information, I would poll my customers to see what they actually use.

  • 1
    I would poll my customers to see what they actually use....Had I realized that was an option for you I would have suggested it long ago! That's a good idea and will give you more relevant data than general population statistics could. – I say Reinstate Monica Feb 24 '18 at 13:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .