mottershead

Feeding the image into Google Images wouldn't help. Requests for an image that come from outside photo.net are filtered out of the counts. Otherwise, the most viewed image might be one that was put on photo.net for the sake of image hosting for a Korean blog or an eBay ad (for example).

 

Regarding the effect of caches, images being in browser or proxy server caches can suppress the view counts. However, many caches do send an "If-Modified-Since" request to the web server on a cached URL, to determine if it is stale. A server will respond "Not Modified" if the URL is still the same, and will send the new content if it has been modified. We count these "If-Modified-Since" requests for the images as "views". This mitigates some, though not all, of the erosion in the counts due to caches.

Peter, regarding your suggestions:

 

1. The forum list only shows items that are "new". Different forums have different definitions of "new". The "New Answers" displays shows any threads that have new answers, including any threads that are regarded "old". If someone replies to a thread from 1998, that thread will appear on "New Answers". By the way, the Minolta forum considers threads "new" for 28 days. Many forums have categories which provide a way to find other posts on particular topics; but the Minolta forum doesn't have categories. This isn't always so useful, since some of the categories get to have a huge number of threads, and just listing them isn't that useful. For these cases, for forums where there aren't any categories, and finally for situations where the pre-defined categories aren't really applicable, there is Google search.

 

2. The "Unanswered" button is currently disabled for the Unified view, but it still works fine in individual forums. In the "unified" context, it is very expensive to generate the "Unanswered" list site-wide.

 

3. This is what New Answers does. I use New Answers frequently, more so than the default forum thread list.

 

4. I like this idea a lot.

It's an election. With no qualifications on the voters other than being able to register on photo.net. Specifically, a valid email address. One person one vote.

 

I keep saying this, but nobody likes this answer. At least not the people who commonly participate in this forum. If your judgement is different than the outcome of the "election", well that just means your judgement is different than the outcome of the election. As if that never happens with elections for public officials.

The way the system currently works, the new members' ratings will get counted the first time the photo's average is computed and the person is no longer considered a new member. That usually is when there is another rating on the photo. If the photo never gets any more ratings, the ratings that weren't counted because at the time they were from new members won't ever be counted. This could be regarded as a bug, but that is how the system currently works.

 

Occasionally, the averages on photos are recomputed by the system en masse. Most often this is the result of a mass deletion of ratings. This might happen if a large number of accounts were discovered to be bogus. It could also happen as the result of some kind of system problem. This doesn't happen very often.

The rating system is designed on the assumption that a single rating does not need to have "credibility". The credibility and legitimacy of the photo ratings as a means of ranking the photos comes from the fact that many people are rating the photos and that each individual photo receives many ratings.

 

Think of it as like an election. No individual voter has to prove his credibility beyond being a citizen and being old enough. The legitimacy of an election is not based on the credibility or wisdom of any individual voter; the legitimacy of the winner comes from being the collective choice. A losing candidate who claimed to be the winner because his voters were smarter, had higher standing, or were somehow more qualified than those of his opponent would be quickly dismissed as an anti-democratic elitist and would get nowhere. The only form of this argument that would get any traction at all would be to question whether some of the voters weren't legally entitled to vote or whether some of the legally-entitled voters had been improperly excluded. But leaving that aside, all votes carry an equal weight. The essence of the photo.net rating system is the same.

 

As for your notion that the low raters "invariably hit with 3/3s before the 4/4s and the rest can react": this is just false. I really must stress that this idea, which is so often repeated in this forum that people now take it for true, is just a flat-out myth. As I reported in another thread, I studied the average of first ratings on photos versus the average of all ratings. There is less than a one-hundredth of a point difference. The rating distribution for first ratings is the same as the rating distribution overall. It just is not true that low raters invariably "hit you" before anybody else. Perhaps it seems that way, but it is not true.

It is a good suggestion and I've thought about this before. It wouldn't be hard to implement. On photo.net just about anything can have a comment on it. The reason I didn't implement it is that folders tend to be a bit volatile. People create and delete folders and move photos around quite a bit. If people were able to comment on folders, they could find either that their comment had been orphaned because the folder was deleted, or the content of the folder could change completely making their comment irrelevant. This could happen with a portfolio comment too, of course. But it is a little less likely. Thanks for the suggestion.

It wasn't only to eliminate complaints.

 

The main reason was that there were exceedingly few 1's and 2's already. This meant that it was pretty much just bad luck whether a photo received one, even assuming it deserved it. To make matters worse, 1's and 2's weren't highly correlated with the average of the ratings that were not 1's and 2's. One would expect a photo with a high average (excluding the 1's and 2's) to have fewer 1's and 2's than a photo with a low average. There was a slight correlation, but not what one would expect. The few 1's and 2's were being sprayed around the Gallery pretty much at random, with high-rated photos almost as likely to get them as low-rated photos. This told me that 1's and 2's were being given for idiosyncratic, or possibly dishonest, reasons. All in all, they were just noise. It could and should have been otherwise, but it wasn't. Once I knew this, it was hard to continue defending the 1's and 2's as legitimate opinion.

 

Given that we didn't need to know which photos were the worst on the site, that they caused unending complaints (even more than the complaints about 3's), and that many of them were probably given for suspect reasons, it was pretty easy to decide that they were more trouble than they were worth.

 

The current scale with 1 and 2 as "virtual" ratings and 3 as the lowest rating that counts doesn't have the problems that the former scale had. Of course, one could argue that 3 could go the same way. as the 1's and 2's, and become a rare and idiosyncratic rating of no use to anybody. Except that it hasn't, and 3 is still a very real rating, more so even than it was.

The view counter counts all hits on any of the two or three sizes of the photo that you uploaded. This includes thumbnails where your photo is displayed with other photos, including your portfolio, your folders and presentations, Top Photos, etc. The view counters are extracted from the web logs, once per day, generally in the early to mid-morning Eastern Time. Your own views of the photo are included.

It isn't an assumption. I already provided evidence. Please see above.

The Featured Portfolio didn't change all this time for a variety of reasons. First, in the early summer I was taking a bit of a break, and only doing the basic things needed to keep the site going. Then a few weeks ago, Philip decided to get involved again and has been asking me to do things on his agenda, not including picking a new Featured Portfolio. Also, as Walter points out, I've been fairly busy lately with technical problems.

 

Finally, at the moment, a new Home Page design is being readied which shows member portfolios in a new way, and doesn't include a Featured Portfolio. So the one on the home page right now will probably be the last Featured Portfolio.

The definition we use is quite reasonable: it is the number of hits from our web logs on the image that you uploaded. You thought it was some other definition, although you didn't say what you thought it was. You asked what it did mean. We told you. Other people have asked the same question, and you could have searched the archives and found the same answer. Where is the deception?

Every hit on the any of the three sizes of the image counts as a view. This includes the thumbnails. So, the view count increases if someone looks at a folder containing the photo, your portfolio, of if the photo is on a Top Photos page.

Lisa, if you were a subscriber, you would be seeing almost no ads at all. #2 is not the current situation. In your options, you forgot #4, which is the real status quo: subscribers pay and see almost no ads, and non-subscribers see ads.

 

The status quo is a blend of your #1 and #3. Effectively, people get to choose whether they pay or see ads. (I say "effectively" because officially people who post frequently are required and expected to subscribe. But, since we leave this to the honor system, many people interpret subscriptions as optional.)

This idea that there are people or robots waiting to slam photos with "instant" 3/3 ratings is really a big myth, a great example of selective memory in operation, and, perhaps, hysteria. One or two people start saying this and pretty soon it is conventional wisdom and everybody is saying it.

 

There is just no basis for this idea. None. Nada. Zilch.

 

During the last week, the average rating given in the "Rate Recent" feature was 4.256 on Aesthetics and 4.196 on Originality. If you look only at the *FIRST* rating that was given to each photo in the "Rate Recent" feature, their averages were 4.245 on Aesthetics and 4.185 on Originality. A hundredth of a point difference between first ratings and all ratings.

 

I don't expect this data to change anybody's opinion, however. When it comes to the evaluation of their own photos, photographers are just not a reality-based community, on the whole -- which is really the basic flaw in the rating system.

Anonymous rating has not been abolished. The Rate Recent interface is causing technical problems and has been disabled temporarily. We are working to reenable it. If you were on the system at the right time today, you would have noticed it briefly re-enabled, in fact. But this brief period showed that the technical problems have not yet been resolved.

Mona, the POW photographers will be asked to agree to the terms of the contest before being considered. If they don't want to make their photo available to HP for advertising, that is entirely up to them. Of course, in that case, they then won't be in the running for the printers. But nobody is going to be presented with a fait accompli.

 

If none of the POW photographers opt in, then I don't know what we'll do with the printers. Right now, they are stacked up in my office. According to Gary Bernstein, the HP 8750 pigment ink jet printer is fabulous. That is what he uses, and is the grand prize during the first two quarters. And for the last two quarters, we'll be awarding a new printer that is coming out in the Fall that is even better than the 8750. Gary, who has seen prints from that one, says it is unbelievable. If none of the POW photographers want them, maybe HP will let me keep them for myself! Although I only really NEED one.

It may have escaped people's notice that at present and for the next year, there are prizes being offered by Hewlett-Packard for having a photo designated as Photo of the Week. The instructions to the Elves regarding how to pick the POW's hasn't changed. It is still intended that the Elves pick photos that they believe merit a week of attention and discussion.

 

But a different set of judges will pick a subset of the POW photographers to receive prizes, based on the overall aesthetic impression and anything else the judges deem relevant. The judges are Gary Bernstein and me. The prizes are very nice Hewlett-Packard ink jet printers.

The system does send one email reminder a few weeks before a subscription is about to expire. We don't want to spam people with naggy reminders, so we only send the one reminder. We also notify people in their Workspace that their subscription has expired.

 

The icon and mail-forwarding address is not removed automatically on the expiration date. We run a procedure to do that about once per month to take away the icons, etc, of subscribers whose subscriptions have expired more than a month ago. It isn't documented that there is a "grace" period, because we don't want people relying on the grace period. We want them to resubscribe. There is no promise of a grace period, and we could change this at any time.

 

The system also sends email notifications to people who receive gift subscriptions.

 

There are many reasons why emails might not get through, and I can't really say why any particular one didn't unless it was very recent and I can track it in the mail logs. Because of all the different schemes mail service providers are coming up with to block spam, email is getting pretty unreliable. I hope this changes before long, because it is big problem for us, but it seems to be taking the email industry a very long time to get its act together regarding spam, and I don't expect this to change in the short-term.

 

Regarding subscribing by credit card from countries that aren't supported by our credit card processor: we currently accept payment from about 35 countries. North America and Western Europe are pretty much covered. Central and Eastern Europe, Africa, Central and South America, Asia, and the Pacific are, with a few exceptions, mostly not covered. If your credit card was issued in one of the countries that is not covered, then you can't use your credit card to pay our subscription fee, I'm afraid. The list of supported countries is expanding (slowly), but that is under the control of the credit card processing company. Unless you have a friend in another country who will do the transaction for you, please continue to participate in the site as a Trial/Guest member.

Everybody usually gets 4-6 weeks of grace to renew before the icons and the photo.net forwarding addresses are removed.

You guys ought to quit discussing the photo rating system if you aren't going to learn a little about statistics. The most common rating of the forest stream photo is 5. (I'm surprised by this because I personally would have given this photo a 6 because of the technical quality and detail.) That is, the most common reaction to this photo was only "Above Average".

 

If the modal rating is 5, it is totally predictable that there will be some ratings above and below that. And this photo does have some above and below. In fact, the distribution is a bit skewed on the high side. But it is totally predictable that there will be 3's on a photo with a modal rating of 5. Indeed, it is a bit surprising to me that there is only one 3/3 on this photo.

 

Turning from statistics to why someone would rate the photo a 3 or 4: that isn't hard to figure out. The 3 and 4 raters didn't notice or didn't care about the technical quality and saw it as a boring blurred stream shot. Streams with blurred water are a cliche subject. There are tons of them on photo.net and other sites. Most people have seen a lot of them. Few of them are this nice, and most people did see the technical quality, rating the photo higher than the typical forest stream photo would merit. But is it so ridiculous that originality is a factor that weighs more heavily with some people than others?

 

Just because you like this photo (as do I) doesn't mean that everybody has to. The one 3 on this shot is hardly a reductio ad absurdum of the rating system.

If looking at photos is like other visual responses, people probably form a judgement in milliseconds. Most ratings on photo.net given in the Rate Recent feature are made in a few seconds, and this time is probably dominated by how long it takes for the photos to download and for people to click the buttons, not by how long it takes for people to form a response to the photos.

Bob is right, those are Google AdWords ads. The advertisers sign up for different keywords like "wedding photography" and Google displays their ads on sites that it believes match those keywords. Only when other people click on those ads does the advertiser pay. The site gets part of that revenue, and Google gets part.

 

If the advertiser didn't choose keywords that match the content and objective of the ads well so that they aren't being displayed to people who are apt to click on them and buy, or if Google didn't profile the pages/sites correctly so that they aren't relevant to the keywords, then Google loses some potential revenue, and the site loses an ad slot that might have displayed a more relevant ad. But the advertiser doesn't pay.

 

Obviously keywords can be very ambiguous. The keyword "Wedding photography" is going to snag sites that are targetted to brides who might want to buy the services of a wedding photographer, which is presumably the audience the advertiser is trying to reach. But it will also match sites like photo.net where wedding photographers are talking shop. I doubt the latter are very interested in the ads or likely to click on them; so those ads are bit of a waste of space for almost everybody. The only person who doesn't care is the advertiser, since he doesn't have to pay. The only time he loses is when a wedding photographer (who is not a prospect) clicks on the ads to check out the competition.

We don't particularly review the new member ratings. We review all the ratings by everybody daily during the week, and sporadically also on weekends. Anyone whose ratings look questionable is going to come under attention. But the problems tend to be from new members. People don't usually suddenly start misbehaving. So suspending ratings is one easy way to reduce the impact of rating misconduct. But there is no process to look specifically at new members and bless them.

Oh, I think I went off on a completely wrong tangent. Maybe the bounce was my ISP (which *is* comcast.net) telling me that the message was bounced at photo.net. I'll get back to you.

Hi, I sent a test message to your terri at photo.net forwarding address, and it did bounce. The bounce message came from the mail server at comcast.net, so it would appear that the photo.net mail server forwarded it correctly. The bounce message was the same as you saw (550 sorry, no mailbox here by that name) referring to the terri at photo.net address . This is weird, because of course there is no terri at photo.net mailbox at comcast. The photo.net mail server was forwarding the message to your comcast mailbox but it looks like the comcast mail server rejected it on the basis of the original To address, which doesn't make sense.

 

To explain a bit more: when mail messages are exchanged between mail servers, there is the message and the "envelope". The "envelope" is the addressing information exchanged between the mail servers, and unfortunately, you don't necessarily get to see the envelope fields in the headers of the message when it is delivered. The To, From, etc headers that you see in your mail reader (Outlook, Eudora, etc) are the headers on the message, not the envelope. The log files of the mail servers may have the only record of the envelope headers. Since the bounce message came from the comcast server, we know at least that the photo.net server was trying to deliver the message to your comcast mailbox.

 

By the way, this is why sometimes people get messages that seem to have been addressed to someone else, particularly with spam. The recipient's address was on the envelope, else it couldn't have gotten into the recipient's mailbox, but the headers on the message itself don't need to match. A message can be delivered even if there is no To: header on the message itself, as long as the envelope is OK.

 

Since I sent the message originally to terri at photo.net, the To: header on the message is that. The mail program on my desktop system delivered the message to my ISP's mail server and this passed the message to the photo.net mail server. When the message was delivered on photo.net, our mail server saw that it needed to be forwarded to your comcast address and sent it on to the comcast mail server. Our server put it in a new envelope with your comcast address as the To: The message inside the envelope was unchanged with the original To: terri at photo.net header. (If you had received the message, this would have allowed you to know that the message had been sent originally to your terri at photo.net address.)

 

It looks like in the process of delivering the message, something on the comcast side opened the envelope, inspected the headers on the message, and bounced the message because the To header on the message was terri at photo.net, rather than a comcast mailbox. This isn't what comcast is supposed to be doing, and it essentially breaks forwarding to your mailbox. Not just through photo.net but through any forwarding service.

 

That is how it looks to me. To be honest, this seems so fundamenentally broken on Comcast's side, that I have a hard time believing that I am interpreting things correctly. Perhaps it is some anti-spam thing gone amok. Anyway, that is how it looks. If any email experts out there want to try sending a message to terri at photo.net and looking at the bounce message that comes back, I'd appreciate a second opinion.