| Pages in topic: [1 2] > | Suggestion: semi-randomise results in directory listings Thread poster: Samuel Murray
|
|---|
Samuel Murray 
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
[Perhaps "randomise" is the wrong word. The results themselves aren't randomised. It is the offset that is random.]
Hello everyone
Here's yet another suggestion about changing the way directory search results are displayed.
Currently, the main factor that determines a member's position in the directory search results is the nu... See more [Perhaps "randomise" is the wrong word. The results themselves aren't randomised. It is the offset that is random.]
Hello everyone
Here's yet another suggestion about changing the way directory search results are displayed.
Currently, the main factor that determines a member's position in the directory search results is the number of KudoZ points. I don't think this is fair to new members or to members who are active participants in ProZ.com but whose main participation is not KudoZ. I also don't think that sorting by KudoZ points it is successful in ensuring that good translators are closer to the top of the list and that bad translators are closer to the bottom of the list.
My suggestion is this: display the results in order of membership number, offset by a random number.
In other words, when a visitor performs a search (after optionally specifying various search criteria), ProZ.com assigns a hidden search criteria in the form of a random number between 0 and the total number of users, and displays the results in order of users' membership number, starting from the member number closest to the random number. This will ensure that the display of results is truly random but still entirely predicable.
(Members with member numbers close to 0 will have a fair chance, like everyone else, because if the random number is so high that it is close to the latest member number, the results will simply flow over to the start of the sequence of member numbers.)
The random number should form part of the search URL so that visitors can bookmark a search result page and later still be able to return to a results page that is practically identical to the one they saw when they created the bookmark for that page.
One can also have a "Save this search" button on the search results that will save the search e.g. as a results URL in a cookie. In fact, it may be a good idea to save a user's previous searches anyway, so that he can return to the exact same result later.
Optionally, ProZ.com can then sort each individual results page by things like KudoZ points, to make sure that those with high KudoZ scores etc still occur at the top of each individual page.
Your thoughts?
Samuel
[Edited at 2013-10-21 13:08 GMT] ▲ Collapse
| | | |
What does randomizing search results even achieve? Please forget the specifics of how for a second; just explain why. What is the big idea, what is the rationale? I just don't get it.
Thank you in advance.
[Edited at 2013-10-21 12:13 GMT]
| | | | Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Mikhail Kropotov wrote:
What does randomizing search results even achieve?
Well, it achieves fairness to everyone in the directory. If results were ordered strictly by member number (without the offset), then people who registered first would be advantaged and people who registered last would be disadvantaged. Currently, the results are ordered by KudoZ points, and the original logic was that this would be fair because it would advantage the most helpful members, but that logic has proved to be false (see below).
Surely all paying members should have an equal chance at being found in the directory searches? By using my suggestion (i.e. using membership number, but not starting at 0), all members have the same chance of being included in any search result.
Or do you believe that some members should always appear higher in the search results, and that other members should always appear lower in the search results?
==
The current listing method (based on KudoZ points) is not fair because it prevents users from appearing in the search results unless they invest a disproportionate amount of effort (i.e. collecting a whole lot of KudoZ points). The original logic with KudoZ was that members who have lots of points will have been proven statistically to know more about a certain topic or subject field, as judged by their peers, but the fact is that KudoZ is too easily manipulated, and besides, KudoZ can only be a reflection of a member's knowledge if that member actually participates in KudoZ.
You are fortunate, Mikhail, that ProZ.com uses the *total* number of KudoZ points for directory ranking, and not e.g. the annual number of points, because if that was the case, you would have been excluded from most directory searches in 2007 (when you accumulated almost no KudoZ points, despite having had 2500 points in total at that time), and would have been included in almost all directory searches in 2009 and 2010 (when you collected an average of 2300 points per year), and you would have been excluded from quite a few directory searches last year (when you collected a mere 200 points). Now just think how unfair it must seem to people who do not participate in KudoZ at all, or who don't collect so many points, if the odds that they are included in the directory results depend entirely on it.
[Edited at 2013-10-21 12:59 GMT]
| | | | Erik Freitag
Germany
Local time: 12:37
Member (2006)
Dutch to German
+ ...
| Not sure that would help, but... | Oct 21, 2013 |
Dear Samuel,
I see the problem, but I'm not sure that randomizing would solve it in a meaningful way.
This discussion comes up every few months again, and always possible solutions are first proposed by peers and then ignored by staff. I doubt this is going to change.
For what it's worth, I'll repeat what I've said in more than one similar threads: A good solution IMHO would be to offer sorting by the ratio of answers selected to questions answered, ideally... See more Dear Samuel,
I see the problem, but I'm not sure that randomizing would solve it in a meaningful way.
This discussion comes up every few months again, and always possible solutions are first proposed by peers and then ignored by staff. I doubt this is going to change.
For what it's worth, I'll repeat what I've said in more than one similar threads: A good solution IMHO would be to offer sorting by the ratio of answers selected to questions answered, ideally with a minimum threshold of questions answered. I'm not exactly sure why proz can't see their way clear to at least make this ratio publicly availabe. ▲ Collapse
| | |
|
|
|
Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
efreitag wrote:
I see the problem, but I'm not sure that randomizing would solve it in a meaningful way.
What makes you say that?
If the problem can be defined as "directory search results always favour specific translators over others", then my suggestion would surely solve that because no-one would always be favoured (with each search, someone else will be favoured). Or do I misunderstand your comment?
A good solution would be to offer sorting by the ratio of answers selected to questions answered...
Your suggestion does not reduce the risk of manipulation (members with enough friends will always have a high ratio) and your suggestion disadvantages people who contribute a lot to KudoZ but whose questions are not frequently chosen (since only one answer out of all correct answers can be chosen). Also, your suggestion is still based on KudoZ particupation -- people who don't participate in KudoZ will always be at the bottom of the list, in your suggestion.
My suggestion will help members who don't collect KudoZ points get an equal chance of being included high in the search results.
| | | |
Samuel Murray wrote: Mikhail Kropotov wrote:
What does randomizing search results even achieve? Well, it achieves fairness to everyone in the directory. If results were ordered strictly by member number (without the offset), then people who registered first would be advantaged and people who registered last would be disadvantaged. Currently, the results are ordered by KudoZ points, and the original logic was that this would be fair because it would advantage the most helpful members, but that logic has proved to be false (see below). Surely all paying members should have an equal chance at being found in the directory searches? By using my suggestion (i.e. using membership number, but not starting at 0), all members have the same chance of being included in any search result. Or do you believe that some members should always appear higher in the search results, and that other members should always appear lower in the search results? == The current listing method (based on KudoZ points) is not fair because it prevents users from appearing in the search results unless they invest a disproportionate amount of effort (i.e. collecting a whole lot of KudoZ points). The original logic with KudoZ was that members who have lots of points will have been proven statistically to know more about a certain topic or subject field, as judged by their peers, but the fact is that KudoZ is too easily manipulated, and besides, KudoZ can only be a reflection of a member's knowledge if that member actually participates in KudoZ. You are fortunate, Mikhail, that ProZ.com uses the *total* number of KudoZ points for directory ranking, and not e.g. the annual number of points, because if that was the case, you would have been excluded from most directory searches in 2007 (when you accumulated almost no KudoZ points, despite having had 2500 points in total at that time), and would have been included in almost all directory searches in 2009 and 2010 (when you collected an average of 2300 points per year), and you would have been excluded from quite a few directory searches last year (when you collected a mere 200 points). Now just think how unfair it must seem to people who do not participate in KudoZ at all, or who don't collect so many points, if the odds that they are included in the directory results depend entirely on it. [Edited at 2013-10-21 12:59 GMT]
Thanks for your reply, Samuel.
I see you are still rehashing the same ideas that have been brought up countless times before. Oh well.
[Edited at 2013-10-21 15:12 GMT]
| | | | Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER | Rehashing an old idea? | Oct 21, 2013 |
Mikhail Kropotov wrote: Samuel Murray wrote:
By using my suggestion (i.e. using membership number, but not starting at 0), all members have the same chance of being included in any search result. I see you are still rehashing the same ideas that have been brought up countless times before.
That is a surprise to me. Sure, I haven't followed all the old threads, but as far as I know the idea that I'm proposing here is entirely new.
| | | | Erik Freitag
Germany
Local time: 12:37
Member (2006)
Dutch to German
+ ...
I really doubt that randomly picking a translator is what potential clients will be looking for.
| | |
|
|
|
| Prioritizing people in searches | Oct 21, 2013 |
Samuel Murray wrote: Mikhail Kropotov wrote: Samuel Murray wrote:
By using my suggestion (i.e. using membership number, but not starting at 0), all members have the same chance of being included in any search result. I see you are still rehashing the same ideas that have been brought up countless times before. That is a surprise to me. Sure, I haven't followed all the old threads, but as far as I know the idea that I'm proposing here is entirely new.
Sorry, I don't have time right now to dig up the relevant links to forum threads nor even to summarize them here, but there have been dozens of them over the years.
But I will voice my opinion on this question you raised:
Or do you believe that some members should always appear higher in the search results, and that other members should always appear lower in the search results?
Of course I think some members should be prioritized over others. The criteria for this may vary, and the threads I mentioned are exactly where such criteria have been proposed and discussed.
| | | | Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER | It's not the picking that's random | Oct 21, 2013 |
efreitag wrote:
I really doubt that randomly picking a translator is what potential clients will be looking for.
My suggestion is not that clients get forced to pick translators at random, but that clients are given the opportunity to see a fair selection of available translators instead of just the translators who happened to have amassed the top 50 number of points.
If you use the directory system by specifying specific search criteria, then you're not picking a translator at random. And if you're browsing through a search result, then you're not picking a translator at random either.
The fact that the list is generated using a random number does not mean that clients' decision about which translator to use would be random. Clients will still choose the best translator in the list. My suggestion is just to increase the size of the pool of translators from which the list is generated.
In a sense the current situation is also "random" in the sense that you use the term, because clients currently have no control over which translators will be chosen to appear in the top spots. In my suggestion, clients will have the exact same amount of control over which translators appear in the first page of the search results.
[Edited at 2013-10-21 15:44 GMT]
| | | | Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Mikhail Kropotov wrote: Samuel Murray wrote:
As far as I know the idea that I'm proposing here is entirely new. I don't have time right now to dig up the relevant links to forum threads nor even to summarize them here, but there have been dozens of them over the years.
I think we're missing each other here: there have been dozens of suggestions to change the way directory results are generated, but I don't think the specific idea of using member numbers (offset by a random number) has been suggested before.
My suggestion is not: let it be changed, but it doesn't matter how. My suggestion is: let it be changed, like this.
Or do you believe that some members should always appear higher in the search results, and that other members should always appear lower in the search results? Of course I think some members should be prioritized over others.
Well, unless I misunderstand your point, then I think we simply disagree on that. I do not believe that a select few members should always be at the top of the listing. There are hundreds of thousands of translators on ProZ.com, but the first results page contains only 50. I cannot agree with a system that allows those 50 spots to be filled by the exact same members every time.
Imagine how it would have been if the search results were simply alphabetical (instead of based on KudoZ). Then those with a username starting with "A" would always be on the first page. Do you really think that that would have been a good thing? I do not.
| | | | Sheila Wilson
Spain
Local time: 11:37
Member (2007)
English
+ ...
| From bad to worse? | Oct 21, 2013 |
Samuel Murray wrote:
members with enough friends will always have a high ratio
But has that ever been proven? I'm not at all sure it has. Of course, those who contribute very often to KudoZ become better known, but that can be for better or for worse. If I agree with or give points to an answer given by someone who consistently gives good answers, who I know and respect through their ProZ.com involvement, then I do so on the basis of yet another good answer, not because I "LIKE" them. This isn't a school playground.
My suggestion will help members who don't collect KudoZ points get an equal chance of being included high in the search results.
OTOH, ranking through KudoZ points has been the system here for many years (since the site was launched? I don't know). It's never been a secret so we knew when we joined, and I suspect it's been a very important factor in ProZ.com becoming the enormous success it is today.
I'd quite like to see more factors taken into account to produce a ranking weighted by more than KudoZ points alone, although it would be complicated, but I really don't see that re-shuffling the pack every time a client does a search is going to be very useful to anyone. Do any of us really want clients to be faced with a truly awful GoogleTranslate user at the top of Page 1, even if only for five minutes? ProZ.com already has to face criticism from various quarters about its image within the industry, and this would surely only serve to exacerbate that situation.
| | |
|
|
|
Samuel Murray wrote:
Imagine how it would have been if the search results were simply alphabetical (instead of based on KudoZ). Then those with a username starting with "A" would always be on the first page. Do you really think that that would have been a good thing? I do not.
See, that's a perfect example of a bad criterion. KudoZ points have their issues, but they work. I learned the rules of the game and played by them. I contributed and continue to contribute to the site in various ways, and the rewards have been worth it.
Also, I don't know about your language pairs and specialization areas, but in mine your statement simply doesn't hold true. It is NOT the same 50 people showing up in search results every time. Well, within a day or maybe a week, it is. But if someone wants to earn a spot at the top, and they are a good translator, then it doesn't take that long to achieve. It's not the total number of points that matters, but the number of points in your specific specialization area(s). And that number is usually quite low: it rarely reaches 3 digits.
| | | | Samuel Murray
Netherlands
Local time: 12:37
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
Sheila Wilson wrote:
OTOH, ranking through KudoZ points has been the system here for many years (since the site was launched? I don't know). It's never been a secret so we knew when we joined, and I suspect it's been a very important factor in ProZ.com becoming the enormous success it is today.
To say that the success of ProZ.com is largely the result of the fact that directory searches are determined by KudoZ ranking is sweeping speculation (and I don't think there is any evidence to suggest it).
Of course I don't disagree that KudoZ ranking has been used for a very long time, and I really don't understand why you think it is relevant whether it is a secret or not a secret. The reason for my suggestion has nothing to do with secrets or non-secrets. It has to do with fairness towards all members and (believe it!) with fairness towards clients.
I really don't see that re-shuffling the pack every time a client does a search is going to be very useful to anyone.
It will be useful to translators who haven't had a chance to be on the first page, purely because they don't participate in KudoZ.
It will be useful to clients, because clients will see a much greater variety of translators than just those who happen to have lots of KudoZ points; and also because clients will more likely see translators who are more available than those who are fully booked due to the fact that they are always at the top of the list and therefore contacted by most clients.
Don't forget that the results will not appear "re-shuffled" from the client's perspective (except for clients who always perform the exact same search every time). Unless the client already knows what the results will look like, there will be no difference for the client, regardless of whether the same translators are always on page 1 or whether entirely different translators are always on page 1.
Do any of us really want clients to be faced with a truly awful GoogleTranslate user at the top of Page 1, even if only for five minutes?
While it is true that translators with lots of KudoZ will likely be good translators, it is patently untrue that translators with no KudoZ points will likely be bad translators. I have discovered this very recently when I visited some of the lower pages of the search results and were surprised to discover the names of several people whom I know to be good translators but who are listed at the bottom of the results page for no other reason than the fact that they don't do KudoZ.
And besides, "to protect the reputation of ProZ.com" should not be the main reason why a certain method is used for determining translators' position in the search results.
| | | |
Has there ever been any statistical analysis done of whether KudoZ points even correlates to translating ability at all? I know that ability isn't that easy to quantify, but it really is at the heart of this issue as I suppose it always has been. If a positive correlation cannot be proven in any way, then I don't see how we can argue that having the directory ranked by KudoZ points is any better than having it sorted alphabetically. And if that is true, then Samuel's suggestion (or some other ra... See more Has there ever been any statistical analysis done of whether KudoZ points even correlates to translating ability at all? I know that ability isn't that easy to quantify, but it really is at the heart of this issue as I suppose it always has been. If a positive correlation cannot be proven in any way, then I don't see how we can argue that having the directory ranked by KudoZ points is any better than having it sorted alphabetically. And if that is true, then Samuel's suggestion (or some other randomizing method) would seem to be the most fair for all users of this site. ▲ Collapse
| | | | | Pages in topic: [1 2] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Suggestion: semi-randomise results in directory listings | Trados Business Manager Lite | Create customer quotes and invoices from within Trados Studio
Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.
More info » |
| | Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
