7

Basically, i need to know how long (on average) and ip address is used per user, this can be US-only stats or World-Wide stats, if you have a source please state it. If its your own data please let us know how you've come to this determination. If its off the top of yoru head (which is fine) please feel free to elaborate on it.

Because some routers stay on for days, weeks and months, this really throws speculative accuracy off as many people i am assuming are holding onto the same IP address for really long amounts of time. In the earlier days most people would have a new IP address when they reboot (dialup etc), but i fear this is no longer the majority.

this information is crucial to me in programming an interface that needs the average amount of time an IP address is used (on average) to determin a few things. Any form of information will help, also, im not interested in an alternative way of doing what im trying to accomplish, which is why i didn't want to get into that detail, but rather, just this time average for anyone who has any idea of it.

also, i tried google. i really could not find the stats, i suspect i don't know how to construct the query for this. also, this information is for non-mobile internet (by this i mean excluding cell phone, ipad, blackberry, iphone etc internet access). also, i am not tracking the person, just the ip address and how long it is in use by any given user until it changes or is unleased. thanks.

Found Suitable Answer:

Here is a suitable answer for anyone interested, if you find this answer useful please feel free to upvote. thanks.

http://blog.comscore.com/2008/09/the_myth_of_static_ip_1.html

This report is written very well and has more than just average information, if you find other suitable reports or sources, please post to comments (since discussion/answering was closed), comments are accepted!

user51047
  • 181
  • 1
  • 6
  • 3
    There is a big difference between an IP being changed and being unleased. Leases are often rather short, like an hour or two, then the user needs a new lease. Very often they get the same IP in the new lease, so a new lease doesn't mean an IP change. – Guffa Sep 08 '10 at 05:44
  • 4
    I can't resist the compulsion to ask you why or how anyone could ever possibly benefit from knowing the mean duration of IP address leases on the internet. Wouldn't you also like to know the median and the standard deviation? Why not guess? –  Sep 08 '10 at 05:46
  • what is a lease? whats the % of people who get a new lease also get an ip off the top of your head? im wondering if anyone has any online stats report or publication on average time ip in use? –  Sep 08 '10 at 05:47
  • ill account for the lease seperately, im just after an average number, which is calculable and reportable by online statistics or surveys of some sort, or reports, which i cannot find. thanks. –  Sep 08 '10 at 05:53
  • why on earth would someone downvote this? am i now allowed to ask for an average on a number that can vary widely? that's just stupid. –  Sep 08 '10 at 05:55
  • 2
    You'll find it hard to find these statistics, because there is no practical use for this information. – Stephen Cleary Sep 08 '10 at 05:56
  • 2
    there is practical use actually, when trying to determin unique visitor counts, when to reset, also when blocking users based on ip, to determine how long this block is meaningful to keep, etc... and many more. –  Sep 08 '10 at 05:59
  • 1
    "Average" isn't a meaningful statistic here. It's liking asking for the average amount of money in someone's pocket. – Seth Sep 08 '10 at 06:34
  • Oooh, ooh, I've got $5.60 (AUD). I'm assuming everyone else of the 230,000 or so other people here has the same amount, so I can base my assumptions on that going forward :-) –  Sep 08 '10 at 06:39
  • 1
    To make concrete the point about the average being meaningless, consider that AOL users have a different IP address for every single web request. –  Sep 08 '10 at 07:32
  • 1
    the fact that your willing to call something meaningless without even knowing the context with which i intend to use this average is staggering to me, you're suppoised to be logical programmers, and your comments are just beyond belief. i understand it becomes less meaningful as the variation increases, but for my application, it is very meaningful. i ask only average since i am certain people won't have any more, since average alone is hard to find. of course, a skewed distribution i would love, hence my question on average or anything one can point to. –  Sep 08 '10 at 07:36
  • if some of you are just here to beat someone up or make sarcastic remarks for your own emotional fulfilment, then i can make some suggestions for other sites, perhaps google "emotional psychological problems help" and that may better suit you. –  Sep 08 '10 at 07:38
  • 6
    @Erx: The fact that you're not willing to *share* the context makes this question less useful than it otherwise would be. If you're going to ask for something which doesn't sound meaningful but where your particular context *makes* it meaningful, you should describe that context. – Jon Skeet Sep 08 '10 at 07:42
  • 1
    @jon, i did not include context because i know how these discussions go, people speak of alternatives when i have considered them all, instead of answering the question, the thread turns into something else. the question itself is a simple question with a simple answer, this is all i want, i do not want alternative suggestions to my application as i know what they are. so it wasn't to be difficult, if people don't know they shouldn't really try to pick on your ability to interpret an average, instead they should say "i dont know" or not comment. –  Sep 08 '10 at 07:54
  • 2
    @Erx, there's a long tradition on SO of people telling other people (based on _available information_) that they should be careful in what they're trying to do, almost as long a tradition of telling people you disagree with that they need psych help :-) _You_ decided to withhold that information (and still do) so you should be little surprised when people try to warn you. You'd get the same warning if you asked how best to write an accounting package in x86 assembler without telling us you want to learn assembler and are very interested in accounting packages. –  Sep 08 '10 at 08:11
  • FWIW, each machine on our work network has a 3-week lease on their IP addresses but, since they're turned on almost every day, the lease never expires. So given that even one of our machines has an infinite lease, the average is ... infinite, no matter what the rest of the machines on the planet do :-) You'll never _see_ those addresses of course, we're in a highly militarised network with all sorts of NAT layers protecting us. –  Sep 08 '10 at 08:15
  • 2
    @Erx: Relax. The SO users who are questioning your question are trying to help you. `One trap that many posters fall into is to ask how to achieve some "small" aim, but never say what the larger aim is. Often the smaller aim is either impossible or rarely a good idea...`, from [writing the perfect SO question](http://tinyurl.com/so-hints). Also see [the Old New Thing](http://blogs.msdn.com/b/oldnewthing/archive/2010/07/28/10043237.aspx) for an example of how this is frustrating to answerers. – Stephen Cleary Sep 08 '10 at 08:35
  • i understand what you are saying, but my question was just the average if anyone knows, and that i did not want any information on alternatives related to method/context, i was clear about that as i know all hte alternatives i did not want people wasting their time trying to discuss alternatives with me. question is just for one average, and if this is not known, then that is ok, you can say you dont know or choose not to comment. –  Sep 08 '10 at 10:03
  • after much searching, i have found this, almost exactly what i am after, and i will try to search for more of these types of reports. http://blog.comscore.com/2008/09/the_myth_of_static_ip_1.html I hope it is useful for someone as there are many people that want to understand the information in the article, also provides more than just an average, and by geography, which is great! –  Sep 08 '10 at 10:05
  • The link to the comScore blog seems to be broken now. The new URL is: https://www.comscore.com/esl/Insights/Blog/The-Myth-of-Static-IP – balu Nov 27 '15 at 13:48

4 Answers4

7

It will likely depend on your audience - and in particular, how many of your users are users of mobile internet. I've found that my IP address on 3G networks changes very often, even though my home IP address doesn't.

You should also take into account users who will access your site (I'm assuming you're creating a web site; you've given us no context) from multiple locations, e.g. home and work.

Ultimately, the best source of information is reality - if you already have some version of the site running, you should measure it yourself.

Jon Skeet
  • 5,134
  • 3
  • 35
  • 41
  • this is why im looking for an average number, since the data can vary so much. also, not interested in mobile, as my interface is windows based.so windows users - it doesn't matter where they connect from etc, just rather just that connection itself (ip) and how long (on average) it stays online. –  Sep 08 '10 at 05:36
  • 3
    @Erx_VB.NExT.Coder: You seem to be assuming that mobile internet = mobile phone. I'm typing this comment on a 3G network, on my Windows netbook. Also, you still haven't defined your target audience - and that could make a *huge* difference to the results. It would *also* help if you'd say what you're hoping to achieve with this. We may have other ideas that could help. Finally, you should define what you mean by IP address here... most users will be behind a NAT. Do you mean the IP address that the outside world sees, or the computer's internal IP address? – Jon Skeet Sep 08 '10 at 05:59
  • hi, IP address that the outside world sees. my target audience are windows users who contume/use this ip address, regardless of their location. of course, im not interested in mobile phone/cell phone users but even if this were in the data then it is better than nothing and i could try to account for it using other means, right now i really have nothing. thanks. –  Sep 08 '10 at 06:03
  • 1
    @Erx: I don't believe your target audience is every Windows user on the planet, or even in the US. Is *every* kind of person equally likely to use your application? – Jon Skeet Sep 08 '10 at 06:16
  • yes, every type of person is equally likely to use my application, and even if they weren't, i know how to account for this and factor it into my data analysis if i needed to. –  Sep 08 '10 at 07:56
2

My estimation: once a month

I live in BC, Canada.

I'm basing this estimate on the fact that I've been working on a project for past month and a bit where I've needed to connect to my local server by explicitly typing my own IP address. It only changed once on me over this period of time.

I shut down my computer every night, so it has to reconnect in the morning.

mpen
  • 11,884
  • 16
  • 55
  • 68
  • thanks mate, i appreciate that. just to confirm this is info based on your connection and yourself alone? or your guestimation on others as well? –  Sep 08 '10 at 05:49
  • 1
    I'm glad you stated it was an estimate, and provided your assumptions. Let me think, that's a sample space of, ... one, yes, that's it, one. Now I'm pretty certain there's a few billion computers on the planet so I'm not _entirely_ certain of the statistical value here :-) –  Sep 08 '10 at 06:37
  • @OP: Yes, just me :) @pax: Hey! He said anything will help. At least I was clear about my sources. If another thousand people report similar findings, then maybe it *will* become part of something of statistical value. Gotta start somewhere, no? – mpen Sep 08 '10 at 07:22
  • @Mark, wasn't trying to have a go at you. Just injecting a little humour :-) –  Sep 08 '10 at 07:32
  • @paxdiablo, i said thank you to him because i am a polite respectful person who ackowledges what is still hte best answer in the room. you can rest assured that i know the value of 1 unit of data from a sample of essentially high millions. you shouldn't assume people would value something in a retarded way, we are all programmers here, not idiots. perhaps you should read the comment i left on my question regarding emotional/psychological help. –  Sep 08 '10 at 07:42
  • @pax: No worries man, I wasn't being serious either :) – mpen Sep 08 '10 at 08:42
  • @Mark, i think he was just picking on me, not you :) –  Sep 08 '10 at 10:08
1

Keep in mind that people connect in so many different ways these days, that it is really hard to come up with exact numbers.

At home I have a static IP, at work I connect through a corporate firewall, which gives me a limited set of IP addresses, when I commute, I use dial-up where addresses are assigned from a large pool and my smart phone can use either dial-up or wifi. In short, as a single user, I have multiple IP addresses with very different life time and I reckon I am not the only one.

You will stand a munch better chance at tracking this via user login than IP.

Brian Rasmussen
  • 236
  • 1
  • 6
  • yes, this is why im looking for an average number, since the data can vary so much. also, not interested in mobile, as my interface is windows based. also, not interested in alternatives, just after a number, which is certainly derivable from some form of online statistics, which i cannot find. –  Sep 08 '10 at 05:44
  • 5
    @Erx_VB.NExT.Coder - but *because* the data can vary so much, an average is both virtually impossible to find, and completely meaningless. – Marc Gravell Sep 08 '10 at 05:49
  • 2
    But "an average number" is practically *meaningless* precisely *because* the data can vary so much. Example: my church has DSL with a static IP; their IP never changes, ever; at home I have DSL and my IP changes every 24 hours. EDIT: Marc is faster than me, but we both said the same thing: the mean (average) is not a [robust statistic](http://en.wikipedia.org/wiki/Robust_statistic). – Stephen Cleary Sep 08 '10 at 05:54
  • i cannot believe you think it is entirely meaningless, i disagree with you there. as it is very meaningful, my audience are windows users, and wide i know, doesn't make it meaningless, as if it is not skewed, then i can estimate more based on a bell curve, but it probably is and ill take that into account seperately. –  Sep 08 '10 at 05:57
  • @Erx_VB.NExT.Coder: Without knowing your target audience, the average across all Windows users *is* pretty meaningless. It could be completely different to the figure your users will actually see. – Jon Skeet Sep 08 '10 at 06:02
  • @Jon, average is what i am starting with, because even that is hard to find and i know its value... i wouldnt' dare ask more as that may alienate people who know of an average or data relating to it. if i could get more, i certainly would, but given average is hard enough to obtain, i presume more would be even harder, so i'd rather ask a question where i have a reasonable probability of an answer than to ask something too obscure & yield no responses. but, typical of a stackoverflow question, i've just gotten beaten up by a dozen people instead. im now understanding the 'average' SO character –  Sep 08 '10 at 07:46
0

I'm no expert, but it depends on the type of connection they have. If they are using dial-up, they would have a different IP every time they reconnect to the internet. This could one or more times per day, all depending on the person. If they have mobile broadband, then it would change even more often. If they have a router which has a continuous connection, then there IP could remain the same for months on end. If they are connecting to the internet from a mobile device, there IP may change as much as 20 times per day.

In short, there is no way to tell. It all depends on your audience, and most people have different connection methods. They could change IP every 10 minutes, or every 10 months.

Alexander Rafferty
  • 69
  • 3
  • yes, this is why im looking for an average number, since the data can vary so much. also, not interested in mobile, as my interface is windows based. –  Sep 08 '10 at 05:39
  • well, I would estimate as an average of all different connection methods: one-three days. –  Sep 08 '10 at 05:41
  • Sorry to ask, but how do you rule out mobile users on account of your interface being Windows based? What about laptops? – Brian Rasmussen Sep 08 '10 at 05:43
  • @Erx_VB.NExT.Coder: Some Windows users have mobile broadband. For example, my home internet connection is through a 3G modem, just like you'd have on a smartphone. – David Brown Sep 08 '10 at 05:44
  • I think he means mobile devices (phone, iPod, iPhone, ect...) not mobile broadband (which I once had). –  Sep 08 '10 at 05:51
  • by mobile i meant mobile phones/cell phones/ipad/iphone etc... i didnt think alex meant laptops as well in the term mobile. but yes, laptops are included. :) oh an mobile broadband is ok as well, since it can service a windows machine :) thanks –  Sep 08 '10 at 05:52
  • @alex: thanx for the figure, i hope it really is 1-3 days average tho... worrying that it might be more, anyone else? –  Sep 08 '10 at 05:52
  • @Erx_VB.NExT.Coder: If you're *worrying* that it might be more, that's a cause for concern. I personally *suspect* it will be more that 1-3 days... but what's the impact on your application if some users have permanent static IP addresses? (Some certainly do here in the UK.) – Jon Skeet Sep 08 '10 at 06:01
  • actually, it is ok either way, i just need an average in order to determine how i will need to change the application. and it is ok that the variation is wide. thanks. –  Sep 08 '10 at 06:09
  • @Erx: If it's okay whatever the value, what does it matter what the average is? Or why not wait until you've released the first version with some arbitrary value, and then tweak it later? Not that I'd say using IP addresses is a good way of tracking unique users to start with... – Jon Skeet Sep 08 '10 at 06:17
  • by "its ok" i mean im willing to put in the work/changes needed which may be necessitated by the difference in the average that i have in mind and the actual/real average... when another user asked "why isn't it ok" that was my answer. just letting you know you've taken my answer out of context. im getting the feeling that most people are just trying to pick at the validity of my question (without grounds to do so) as opposed to help answer it, by do SO users _always_ like doing this? –  Sep 08 '10 at 07:51
  • PLEASE PLEASE PLEASE. Tell me what your application does, and why this IP thing is such a major issue. I'm so curious. –  Sep 08 '10 at 08:04