Photos selected as "Popular" don't actuallyseem to be ones that are viewed that much?

zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
edited July 31, 2014 in SmugMug Support
Not sure if this is a bug or user error on my part but....

I have been trying to add a "popular photos" function to my site. The simplest option, which I have done, is to add a link to http://zubenelgenubi.smugmug.com/popular/. I have also created a smart gallery and populated it using the "popular" filter. (Curiously, the latter gallery only ever contains about 30 photos, even though I set its max photos value to 500. Not sure why this is?)

Anyway, in BOTH cases, one of the photos that is selected near the top of the popularity list is this one:

http://zubenelgenubi.smugmug.com/Events/Ice-Art-2014/i-txBPhGR/0/O/2014_03_16_449_IceArt_Part2.jpg

which surprised me, because I wouldn't have guessed this would be especially popular. So used the statistics pages to check how many times its been viewed - the answer is a total of 24 times since I uploaded it earlier this year. However, according to the statistics, the most viewed photo in that same gallery has been viewed 561 times. So why on Earth wouldn't that one be chosen for the "popular photos" instead of the one with only 24 views? Does the popular photos filter factor in a bias toward recent views only, perhaps?

Here's another photo that made it into the "top 30" gallery of popular photos:

http://zubenelgenubi.smugmug.com/SpacePhysics/ScanningDopplerImagers/HAARPSDI/i-55J4pMf/0/O/DSC_0149.jpg

It's only been viewed 51 times, whereas the most viewed photo in that gallery has 109 views, but isn't in the popular gallery. And NONE of the photos in this gallery have been viewed a lot compared to other galleries.

As another example of strangeness in the popular photos, one of the most viewed single photos on my site is the first photo from my 2013 Equinox marathon gallery. (I can't get a link to it right now because of a known bug in the way smugmug displays large galleries!) Suffice it to say this photo has been viewed 1356 times, but doesn't appear at all in my "top 30 most popular" gallery. (It may appear eventually in the much deeper list at http://zubenelgenubi.smugmug.com/popular. But, if so, it's certainly nowhere near the top.)

Something very strange must be happening with the way smugmug decides which photos are "popular"? Or else perhaps I'm doing something wrong?

- Mark

Comments

  • Tan68Tan68 Registered Users Posts: 9 Beginner grinner
    edited July 9, 2014
    Do 'popular' act on views or 'likes' or whatever those actions are called..?

    If 'liking' social stuff is turned off, I guess 'popular' is defined by views.

    If social linking and whatever is turned on, maybe 'popular' = 'likes'
    Defining 'popular' by 'likes' makes sense if a person is into the social media stuff.

    Does seem odd with the large disparity in views...
    But maybe the images that have been viewed a lot didn't get 'liked' (people looked and moved on) yet the one viewed much less got a bunch of 'like' ...? I don't mean to say the people didn't enjoy the image that has been viewed most. Perhaps they were browsing or doing a slide show and didn't stop to do the 'like' thing.
  • AllenAllen Registered Users Posts: 10,013 Major grins
    edited July 9, 2014
    As I see it the Smugmug "Like" has nothing to do with social media, it's only there to replace the "thumbs up"
    that legacy had. Popular photos use this "like" plus comments and some other mysterious things. I don't think
    "views" is even considered.

    I think it's very misleading as almost everyone thinks this is a Facebook "like" as it is almost everywhere now
    on the web.
    Al - Just a volunteer here having fun
    My Website index | My Blog
  • livverlipslivverlips Registered Users Posts: 31 Big grins
    edited July 9, 2014
    zubenelgenubi as of July 1st logged in users now count towards image views in smugmugs stats. Basically you cannot use stats reliably as you have in the past any longer. See my other posts and threads for more info on that if you wish. But no, it is my understanding that the "popular" photos were determined by the thumbs up option as others have stated.
  • mbonocorembonocore Registered Users Posts: 2,299 Major grins
    edited July 9, 2014
    livverlips wrote: »
    zubenelgenubi as of July 1st logged in users now count towards image views in smugmugs stats. Basically you cannot use stats reliably as you have in the past any longer. See my other posts and threads for more info on that if you wish. But no, it is my understanding that the "popular" photos were determined by the thumbs up option as others have stated.

    That would only be your initial view though, so that shouldn't make even be considered when the photo has hundreds of views.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 10, 2014
    Thanks everyone for the clarifications here.

    FYI, I also sent this same question to the Support Heroes, and they basically confirmed what has been explained here: the only data used to determine popularity is number of presses of Smugmug's own "thumbs up" button. Presumably this makes excellent sense in many or possibly most cases - but unfortunately not in mine.

    Only a miniscule fraction of the photo views on my particular site end with a click of the thumbs up button. Either nobody likes any of my photos, they can't be bothered clicking the like button, or they never even knew it was there. Well, probably it's some combination of all three. There just aren't enough clicks of the button to produce a meaningful selection of photos. Those that are selected this way certainly aren't representative of the best content on my site, in my opinion. But, by contrast, there certainly IS a statistically meaningful number of photo views - typically at least a few hundred per day and, on occasion, many more than this. Using these data WOULD select the photos that have attracted most attention. (Caveat: Featured images for galleries always attract most attention, but that's okay.) Anyway, I don't imagine that Google would fail to leverage the information contained in their visitor statistics!

    I have no idea how Smugmug is implemented. But I do know enough about data analysis that I can't believe it would be hard to offer a new filter option for smart galleries - "most viewed". This would in no way replace or usurp the existing algorithm for selecting "popular" photos; it would simply be another option for populating a smart gallery, with a different and immediately comprehensible name. I see no down side to offering this option, other than the effort to implement it - and I would be surprised if that was a lot.

    Still, I also imagine the developers at smugmug are inundated with heaps of equally worthy feature requests. I am still very fond of Smugmug, and I appreciate that they can only do so much. So, for now, I'll probably abandon any attempt to display "popular" photos, until either Smugmug implements some additional options, or some Tremendously Clever dgrinner figures out a hack that allows us to do it for ourselves. Thanks again for the feedback!

    - Mark
    mbonocore wrote: »
    That would only be your initial view though, so that shouldn't make even be considered when the photo has hundreds of views.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 10, 2014
    Thanks for this "heads up" on changes to how statistics are calculated
    livverlips - thanks for this clarification. I had noticed recently that there was a suspicious correlation between my photo viewing statistics and my own activity on my site, so I had begun to suspect something was not right. I'm glad that you've confirmed this.

    This new behavior seems extraordinarily unhelpful to me. But presumably there really is some quite sensible rationale for it that I'm just not seeing? I would love to know what it is! I wonder if smugmug could give us the OPTION to choose for ourselves how we want our stats to be compiled?

    - Mark
    livverlips wrote: »
    zubenelgenubi as of July 1st logged in users now count towards image views in smugmugs stats. Basically you cannot use stats reliably as you have in the past any longer. See my other posts and threads for more info on that if you wish. But no, it is my understanding that the "popular" photos were determined by the thumbs up option as others have stated.
  • denisegoldbergdenisegoldberg Administrators Posts: 14,385 moderator
    edited July 10, 2014
    This new behavior seems extraordinarily unhelpful to me. But presumably there really is some quite sensible rationale for it that I'm just not seeing? I would love to know what it is!
    Baldy presented this in the first post of the thread Upcoming change that will affect stats.

    --- Denise
  • Tan68Tan68 Registered Users Posts: 9 Beginner grinner
    edited July 10, 2014
    If it doesn't have the little blue icon and/or doesn't say 'like me on Facebook', then it isn't a FB 'like' situation. But it is still social media. 'Social media' doesn't mean only Facebook. Any system that incorporates thumbs or likes or purple teddy bears to indicate approval is social media. It kinda defines itself... people socializing. liking stuff. swimming pools, movie stars. dennis hopper. etc.

    It is nice to take advantage of the 'like' system if the person is into social media. Sounds like the OP would prefer to go with 'views'. I think I would agree with him, here. As I noted before, 'likes' (FB or not) tell a story. But the story written depends on the audience. The OP may have an audience that really isn't into social stuff. His visitors look but don't like yet that doesn't mean they didn't enjoy. It would be nice to be able to toggle between popularity based on looks rather than likes. Nice option to have if it is possible to work out.

    Edit > as an example of audience:
    Say a photographer has a site full of landscapes and most people look at these. They are just nice landscapes and attract a crowd of people that just enjoy simple landscapes. These people may not really be into 'liking' things. They just view things. The photographer also has a few images he took when he was trying HDR and super contrasty street images (a phase). These images don't get many views because most of his visitors prefer the nice landscapes. These gritty street images do get a few views from a different audience and this audience is tuned in with 'liking' so they click the button. These images, although they really aren't popular, float to the top... because they have more 'likes'. A thing like this could happen...
  • livverlipslivverlips Registered Users Posts: 31 Big grins
    edited July 11, 2014
    mbonocore wrote: »
    That would only be your initial view though, so that shouldn't make even be considered when the photo has hundreds of views.

    mbonocore, I assume that you saw my gallery that I PMed you about? I infact had that gallery private with a password and I had a total of over 700 image views, up to 17 or more views for each image size in an 80 or so image gallery and they were almost all from me. This was over a 48 hour period. It is possible that I had cleared my cache once maybe twice over the course of my editing the gallery but if you're statement is accurate I would have only had 3 views per image (size) at the most. That was clearly not the case. Since I haven't heard back from you about my PM let me know if you were able to find any more info about what you saw in that gallery, or if you have any more info on the views. Maybe there was some other way the images were being seen? I may do some more testing on my own over the weekend.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 11, 2014
    I see the same behavior - whether or not my browser (Firefox/Win7) caches photos or re-fetches them is highly sporadic and unpredictable - but it is never just one fetch with the cache used exclusively thereafter. Since the change in the way stats are compiled I have seen a jump of perhaps 400% in my site's total photo views per day (~500/day jumped up to several thousand per day). I have been tinkering a lot, and the increased volume correlates pretty closely with how active I was on any given day. So, at least in my case, the new behavior pretty much does mean that "user" statistics are entirely buried by my own use. Shame really.

    - Mark
    livverlips wrote: »
    mbonocore, I assume that you saw my gallery that I PMed you about? I infact had that gallery private with a password and I had a total of over 700 image views, up to 17 or more views for each image size in an 80 or so image gallery and they were almost all from me. This was over a 48 hour period. It is possible that I had cleared my cache once maybe twice over the course of my editing the gallery but if you're statement is accurate I would have only had 3 views per image (size) at the most. That was clearly not the case. Since I haven't heard back from you about my PM let me know if you were able to find any more info about what you saw in that gallery, or if you have any more info on the views. Maybe there was some other way the images were being seen? I may do some more testing on my own over the weekend.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 11, 2014
    Tan68:

    Your example scenario illustrating why a "like" button can be misleading is very relevant to my situation. In my case I've taken a few photo shoots at local sporting events (marathons & races.) None of these photos are very good; I was just trying to learn something about how to take sport photos. However they do attract a disproportionate fraction of the (rather few) "likes" on my site. This is NOT because they're any good. Rather, it's the inevitable consequence of human behavior - people like seeing photos of themselves.

    I have no objection to Smugmug's current method of basing popularity on the frequency of likes that a photo attracts. However, if I were in charge (tee hee hee....) I'd also offer other ways for us to determine popularity. What I'd actually do is try to implement some fancy scheme in which each user could individually tweak the popularity algorithm to set how much relative weight should be assigned to a number of potential indicators including likes, photo sales, photos views, and recency. Assuming the underlying data are readily available then designing a flexible popularity ranking algorithm like this would be a piece of cake. But, as we have just seen with the issue now affecting site usage statistics, the availability of the required data might not be guaranteed.

    - Mark

    Tan68 wrote: »
    If it doesn't have the little blue icon and/or doesn't say 'like me on Facebook', then it isn't a FB 'like' situation. But it is still social media. 'Social media' doesn't mean only Facebook. Any system that incorporates thumbs or likes or purple teddy bears to indicate approval is social media. It kinda defines itself... people socializing. liking stuff. swimming pools, movie stars. dennis hopper. etc.

    It is nice to take advantage of the 'like' system if the person is into social media. Sounds like the OP would prefer to go with 'views'. I think I would agree with him, here. As I noted before, 'likes' (FB or not) tell a story. But the story written depends on the audience. The OP may have an audience that really isn't into social stuff. His visitors look but don't like yet that doesn't mean they didn't enjoy. It would be nice to be able to toggle between popularity based on looks rather than likes. Nice option to have if it is possible to work out.

    Edit > as an example of audience:
    Say a photographer has a site full of landscapes and most people look at these. They are just nice landscapes and attract a crowd of people that just enjoy simple landscapes. These people may not really be into 'liking' things. They just view things. The photographer also has a few images he took when he was trying HDR and super contrasty street images (a phase). These images don't get many views because most of his visitors prefer the nice landscapes. These gritty street images do get a few views from a different audience and this audience is tuned in with 'liking' so they click the button. These images, although they really aren't popular, float to the top... because they have more 'likes'. A thing like this could happen...
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 11, 2014
    Thanks Denise - I read Baldy's post and I agree that if it comes down to a tradeoff, the core performance of the site trumps the need for for accurate statistics. Mind you, I'm kinda dumbstruck to think that Smugmug couldn't work with their provider to fix the issue of being unable to verify if the user is logged in? Seems like that would be a critical capability for pretty much any e-commerce traffic?

    Well, it's all very disappointing - but I guess we're stuck with it for now.

    - Mark
    Baldy presented this in the first post of the thread Upcoming change that will affect stats.

    --- Denise
  • mbonocorembonocore Registered Users Posts: 2,299 Major grins
    edited July 12, 2014
    Hey guys,

    Just an FYI, I have a conversation going with my Stats developer and hoping to get some more detailed information as to how/when stats are counted. I personally haven't tested this myself, nor do I ever look at my stats, so I can't really give a good explanation, but I am working on it for you.

    Will update you all as soon as I can, most likely sometime on Monday afternoon/evening.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 13, 2014
    Fantastic, thanks! Being a minor stats nerd, I for one appreciate all efforts to ensure that they're meaningful.

    - Mark
    mbonocore wrote: »
    Hey guys,

    Just an FYI, I have a conversation going with my Stats developer and hoping to get some more detailed information as to how/when stats are counted. I personally haven't tested this myself, nor do I ever look at my stats, so I can't really give a good explanation, but I am working on it for you.

    Will update you all as soon as I can, most likely sometime on Monday afternoon/evening.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 22, 2014
    mbonocore wrote: »
    Hey guys,

    Just an FYI, I have a conversation going with my Stats developer and hoping to get some more detailed information as to how/when stats are counted. I personally haven't tested this myself, nor do I ever look at my stats, so I can't really give a good explanation, but I am working on it for you.

    Will update you all as soon as I can, most likely sometime on Monday afternoon/evening.

    Just wondering if there's been any update on how the stats are counted? I did notice when I looked at the HTML sent to my browser for my smugmug homepage there is a statement near the top that reads as follows:
    <meta http-equiv="cache-control" content="no-cache, no-store, must-revalidate">I'm no expert on how browsers handle caching.... but this sure looks to me like smugmug is for some reason telling my browser not to cache anything. This would be consistent with the behavior I've seen with my browser, in which caching of smugmug content doesn't seem to occur much, if at all.

    - Mark
  • AceCo55AceCo55 Registered Users Posts: 950 Major grins
    edited July 22, 2014
    I'm also hanging out for an answer.
    Stats are pretty important for me - checked several times per week
    I just want the photos in my galleries counted.
    I deleted my home-page slideshows - I was sick of seeing huge numbers for those.
    Also a bit sick of having my logo score a zillion hits and my "Smugmug site files (Do not delete)" - 1300 "views" in the last week.
    As I said - really interested in the response.
    My opinion does not necessarily make it true. What you do with my opinion is entirely up to you.
    www.acecootephotography.com
  • AceCo55AceCo55 Registered Users Posts: 950 Major grins
    edited July 25, 2014
    mbonocore wrote: »
    Hey guys,

    Just an FYI, I have a conversation going with my Stats developer and hoping to get some more detailed information as to how/when stats are counted. I personally haven't tested this myself, nor do I ever look at my stats, so I can't really give a good explanation, but I am working on it for you.

    Will update you all as soon as I can, most likely sometime on Monday afternoon/evening.

    Any news on this please? ... or have I got the wrong Monday?
    My opinion does not necessarily make it true. What you do with my opinion is entirely up to you.
    www.acecootephotography.com
  • AperturePlusAperturePlus Registered Users Posts: 374 Major grins
    edited July 26, 2014
    I am also hanging in for an answer.
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 27, 2014
    It looks to me like "owner views" contribute enormously to photo view stats now
    Just an update on the popularity/photo view stats issue. I have tweaking a number of aspects of my site over the last few days to try to increase my photos' popularity from their current level of almost zero. It's impossible to know for sure, but it really does look to me like I can't do much at all on the site now without producing a jump of at least a few hundred in the viewing stats. (And if I'm working on it for a while it looks like I produce many thousands of hits.) I suspect this is because most of my galleries are now set to use the "collage landscape" style, so merely accessing a gallery means ALL photos in it are displayed (albeit only as small tiles.) Many of my galleries have tens to hundreds of shots in them so, if this assumption is accurate, then I pretty much can't ever look at my site without adding a hefty (but erroneous) bump to my stats.

    Now of course I have no way of actually knowing whether my suspicions are correct. Maybe there really was substantial outside traffic that roughly coincided with the times when I was working? (Unlikely...but not impossible.) I've spent quite a lot of hours over the last few days adding keywords and captions to photos, because I noticed right now my site is essentially invisible to google searches. I'm not optimistic this will help, based on others' experience - but how will I ever know? It seems like the only way to get a reliable measure of outside traffic is for me to stay entirely off my site altogether. And since I get the impression it takes weeks before any SEO tweaks on a site are reflected as changes in traffic volume, this would suggest I have to remain idle for many weeks. That's a non-starter for me, so right now I'm rather at a loss regarding how to understand how well my site is working.

    Any clear and definitive information from smugmug on the real meaning of the stats would at least save me from basing my efforts on blind guesswork!

    Thanks,

    - Mark
  • denisegoldbergdenisegoldberg Administrators Posts: 14,385 moderator
    edited July 27, 2014
    Any clear and definitive information from smugmug on the real meaning of the stats would at least save me from basing my efforts on blind guesswork!

    Per the thread Upcoming change that will affect stats owner views are now included in
    the stats (as of 25 June 2014).

    --- Denise
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 28, 2014
    Per the thread Upcoming change that will affect stats owner views are now included in
    the stats (as of 25 June 2014).

    --- Denise

    Thanks Denise,

    Yes, understood. But the devil is in the details:

    - If I open a "collage landscape" gallery, does every thumbnail that appears add one to my count of photo views? (It looks like the answer might be "yes"?)
    - If I then close the gallery and re-open it, are the thumbnails all counted again? (It looks like the answer might again be "yes"?)
    - If I view the gallery in the organizer, do each of the (now very small) thumbnails add to the photo view count?
    - If step back and forward a few times through a few photos in the light box, does each re-display of an image add one to the count, even if it's already been displayed quite recently? (I presume the answer is "no", although the html for my galleries does seem to explicitly tell my browser not to cache anything.)
    - I presume each image shown on a home page slide show adds one to the counter?
    - I also presume each image shown on a slide show embedded into a page on an external site also adds one?

    The reason I'm worrying is that the exact answers to these questions is that they do make a BIG difference. If every thumbnail gets counted, there's no way I can ever manage my site without obliterating any hope of obtaining useful guidance from the statistics.

    I still really like smugmug ... but I'd like it even more if I felt I could trust the stats!

    - Mark
  • AceCo55AceCo55 Registered Users Posts: 950 Major grins
    edited July 30, 2014
    You will probably not like the latest news on statistics: http://www.dgrin.com/showthread.php?t=248577
    As Smugmug prefetches pages/images, it counts them in the statistics ... even if the viewer ends up not looking at them!!

    So Smugmug is going to be counting images that have not been viewed!!!!!!!!!!!!!!!! What the .... ????
    My opinion does not necessarily make it true. What you do with my opinion is entirely up to you.
    www.acecootephotography.com
  • livverlipslivverlips Registered Users Posts: 31 Big grins
    edited July 30, 2014
    AceCo55 wrote: »
    You will probably not like the latest news on statistics: http://www.dgrin.com/showthread.php?t=248577
    As Smugmug prefetches pages/images, it counts them in the statistics ... even if the viewer ends up not looking at them!!

    So Smugmug is going to be counting images that have not been viewed!!!!!!!!!!!!!!!! What the .... ????

    At least this answers some of my questions. I knew stats were useless now but I didn't know they were made doubly so by Counting page owner views AS WELL as prefetching non viewed images! This is a seriously huge clusterfu..!
  • zubenelgenubizubenelgenubi Registered Users Posts: 67 Big grins
    edited July 31, 2014
    Wow, I read that thread and it is *really* troubling.

    I do understand the need to optimize performance and, reportedly, this change did produce some improvement - although the truth is that I personally didn't even notice it. (Maybe if I had a super fast internet connection I would have, but here in Alaska I presume the limiting factor is my connection, not the site itself.) But I sure did notice that the stats suddenly went berserk!

    Okay, I'm just an enthusiast. So for me the price of losing the ability to tell how popular my photos are isn't any more serious than simply wrecking the fun of posting them in the first place. But if I was trying to run a business, this would be a non-survivable catastrophe. If google has taught us anything, it is how information feeds success. I simply don't see how anyone can run a successful photo business today without accurate knowledge of what their audience finds engaging?

    I sure hope that addressing issue is being tackled by smugmug with the HIGHEST priority....

    - Mark
    AceCo55 wrote: »
    You will probably not like the latest news on statistics: http://www.dgrin.com/showthread.php?t=248577
    As Smugmug prefetches pages/images, it counts them in the statistics ... even if the viewer ends up not looking at them!!

    So Smugmug is going to be counting images that have not been viewed!!!!!!!!!!!!!!!! What the .... ????
Sign In or Register to comment.