Dear Pantheon,

I think your concerns are well founded. An additional problem is that all too many profiles are still cluttered with data from the likes of IMDb instead of actual film credit data. This creates false results when it comes to establishing the most commonly credited name. And then there is the issue of how to count: per profile or per film.

A separate, but related problem is the issue of birth years (both issues have to do with linking names), including the fact that if not for everyone sharing the same name a birth year can be found, restoring a backup will lead to uncontrolled attachment of birth years.

The question then is: how to proceed? Your answer is a drastic one, but I think it deserves consideration. At this point, I personally have no fixed opinion as of yet.

Cheers,
DJ

I have stated in the past that I didn't think this was the best solution. I personally agree with Hal on this one... it should be completely redone as more of an alias system then a credited as system. As he said it would not be a perfect solution... but I do believe it would be better then what we have now.

That said... I will continue to do it as Ken prefers... I just hope he reconsiders it  As it really don't seem like it is working as planned. And as is we will have to continuously revisit a lot of names as the common name goes back and forth.

I am in qualified agreement with Pantheon. Any system that is user driven in selection of the Common is going to be problematic. We also have users that don't understand that CA is feature driven by As Credited, they think that CA drives the car. There should not be a priority name,, it should as I have said several times be a Simple Association method.

The existing priority name system functions as A=B and C and D and E, the correct answer is totally dependent on ALL users entering data As Credited in order to determine the correct answer for A.

AS system that would function A=B=C=D would be much more useful. under this system any search would yield the SAME results. The Search results could be setup to list the searched for name first with the rest in descending order. Or even in any order a user wishes.

It is important that this data is sharable via Contribution, however as I envision it it would not be a system that would even need to be voted on per se. We are simply sharing the associations so that we know that Lon Chaney=Lon Chaney, Jr.= Lon Chaney Jr, the association system has weakness in some areas, for example Lon Chaney could be Jr OR Sr since Jr dropped the usage of the suffix many times after his father passed away. However, this weakness is not major since while it does happen, it is not frequent, in fact of the top of my head I can only think of four such actors involving two name sest. The simple association would also el9iminate the need that some users repeatedly express to standardize such things as suffix handling.

Whatever sort of Alias system is used is important, but users do need to understand that it is a FEATURE that currently is driven by correctness of the basic data entry of As Credited, there are users who refuse to follow the Rules and are still managing to get IMDb data past eveyone, which of course is going to skew our results in some cases badly.

Okay enough of my soapbox.

Skip

Quoting T!M:

Quote:
The worst thing is that accepted data doesn't propagate in any way: if I document a certain "common name", and submit it into, say, the Australian profile for the one DVD in which that actor appears in my database, then nobody else gets to know about it. In every other locality someone has to pick up on the need for a "common name", and do the work all over again.

Off the top of my head:

Even if we'll never get a sepeartion of movie and DVD, it might still be a good idea to somehow identify and "link" DVDs, that contain the same movie (via original title for example).

For example. Let's say Courtney Cox's common name is now Courtney Cox Arquette.

If I now edit and contribute my edition of "Masters of the Universe" and change "Courtney Cox" to "Courtney Cox Arquette, credited as Courtney Cox", then the system could check every other DVD with the same original title and look for the same actress/role. If it exists, it creates a "system contribution" that makes the same changes I did on the other profiles and opens it for voting.


That being said, I exclusively use IMDb data and my linking system works just fine.

I am not happy with the credited as feature either, mainly because I feel the CLT is not a sufficient way to link individuals. I propose to link cast and crew with different appearances (and ONLY them) via their birth name.

Good point, Ken!

Quoting Pantheon:

Quote:
Until recently it seemed to be ok (endless arguments about what constitutes the ‘common name’ aside); but I have noticed contributions coming through that are a little worrying.
I can’t cite specific contributions (I didn’t make a note....sorry!) but I remember the gist of the contribution notes:
‘Until recently (insert actor) has been known as (insert actor) in the Credit Lookup Tool. However, recently this has changed and now the CLT is giving a different result. Subsequently I am changing the Credited As entry to (insert name); which is now the most common entry.’

This is not an isolated incident. I have seen at least half a dozen of these appearing.

It's not an "incident". That's the way the system is designed, for right or wrong. The definition of the name is the most frequently credited name. If someone has 2 film credits and then changes their name and then makes 3 films with the new name, their common name changes. Also, since the db is so populated with IMDb data, using the CLT results now gives you one result. As more "as credited" names are contributed, those common names could change as figures shift. This is not a result of users changing their mind. This is a by-product of how the system is designed. Users looking at CLT data today will be required to choose Name A for example whereas users looking at CLT data a year from now may be required to choose Name B, if figures change the results. It's not their whim, but rather the system.

Quote:
Here’s where I think the problem lies:

1. Not everyone is bothered about the Credited As feature – so they simply copy the cast/crew from the credits.

Submitting "as credited" data refines the CLT results and isn't a "problem".

Quote:
2. Not everyone is bothering to check if there is an ‘established’ common name agreed (whether in the forum thread or the CLT).

It would be nice if users contributed common names and as-credited names together, but the CLT is still improved if as-credited names are contributed.

Quote:
3. Some contributors are simply going by their local database to establish how they personally want to see a specific actor credited.

I don't see this happening. I see contributions which list CLT results as required by the rules. In some situations, users will show that the CLT data is skewed by incorrect data that isn't as-credited, but again, this is not based on their preference but rather it's based on their research into the true as-credited data. I don't see this as a problem.

Quote:
4. These profiles are subsequently submitted which then alters the result of the Credit Lookup Tool – leading to existing Credited As entries no longer being valid.

Again, this is the nature of the system. The system established the DVD Profiler data as the source for the common name, yet the data is still, by my estimation, about 80% from IMDb. As true as-credited data is contributed, the CLT results will change for some names.

Quote:
I am a firm supporter of keeping this feature locally. Here’s why:
1. No more arguments over common names.

There aren't really arguments over common names to the extent that there were before the CLT was established. The CLT results are used by contributors unless someone has research which shows the CLT results are skewed by bad data. The most frequently credited name is the common name. There's not much to argue about there.

Quote:
2. Cast/Crew taken from credits as before.

Cast/Crew is still taken from credits. There's been no change to that.

Quote:
3. No changes to local database by someone else’s interpretation of common name.

Again, these are not interpretations of common names. They're based on CLT results or based on research that shows the CLT results to be skewed. I've never seen a common name get through based on someone saying it's the name they've chosen for their local db.

Quote:
4. No need to revisit audited profiles because a ‘new’ common name has been established.

Yes, that's the nature of the system. There is no fixed common name. For most names, it works fine. But as the CLT data is improved, it can result in changes to previously determined names.

Quote:
I already realise that the majority of people will vehemently disagree with my preference for keeping this feature local only. Fair enough. I just want to express my concerns.

Name linking is one of the best features of the program. I would be surprised to see that feature removed or restricted from contributions.

That's a heckuva long defense, but it doesn't change the fact that the system is...imperfect...to be as polite as I can.

As an active contributor, I can attest to the fact that I only contribute maybe 20% of the cast/crew changes that I make locally as a direct result of this policy.  It's arbitrary and leads to seesawing of names, potentially with no end in sight given changing marital status.  If I'm feeling particularly generous, I'll go ahead and break my profile and just submit the straight credits and then immediately go back and fix my local profile.

I've given up any real hope of this ever effectively getting cleaned up in the online and have instituted my own set of rules for my local.