Quoting Unicus69:
Quote:
What we have here are two sided of the same coin, each ignoring the piece of data they want to ignore. One side wants to ignore 'titles' while the other wants to ignore 'profiles'. My question is this...if we were meant to ignore one in favor of the other, why were we given both?
On the "edge" of that coin, some of us would like to count only actual "film credits" and ignore both "titles" and "profiles." My thought as to why the CLT results are presented as we now have them; it was the simplest way that Ken thought of when he was making the tool. The rules are definitely not clear, however, Ken did state we were to use the most credited form. Since film credits, for the most part, do not change from region to region, across editions, or even in re-release, actual credits should be the determining factor when ever someone bothers to dig them up. In the rare instances where actual credits are different, that single title would self negate since it would be adding 1 to the count for each variant actually credited whether it is only 2 variants or 200.
As has been said before, the
most credited form can not mean profiles, because profiles are not credits, especially with the preponderance of IMDB data currently residing in a lot of profiles. 2 profiles for Chocolate have IMDB type roles:
Madame Pouget (as Dominique Mac'Avoy). If all of the profiles were accurate representations of the credits, the profile count would be more indicative, but would still be inaccurate for the most credited form if we take those words literally. Of the 51 total profiles for the actress in question, 43 profiles from a single film (read 1 "credit") make Mac'avoy count as 1 with 2 film credits each for the other variants. Of the other 2 names, Mac Avoy has 2 titles 2 profiles, and MacAvoy has 2 titles 6 profiles. Using profiles as the tie-breaker, MacAvoy would be the winner provided the actual credits are matching the CLT for those 4 films.
The problem with using the CLT for Titles, is we do not get an accurate count without manually sifting the results. Looking at the results in the first post, the CLT erroneously reports a total of 13 titles where, in fact, there are actually only 5. The profile count is more "stable" since that number does not change like the "title" count can, but like this particular actress, one film of the 5 is skewing the results.
However, for someone who has a more prolific career, with 40, 80, or hundreds of "CLT titles", this is a very cumbersome process. Miguel A. Nunez, Jr. has 193 "titles" and 371 profiles according to the summation of CLT results for all 12 variants that can be found.
I realize that "correcting" CLT results by simple subtracting and adding is wrong, because the subtracted title/profile count might not equal the added result, due to the method of "title" counting used by the CLT. There are so many polluted profiles with the IMDB name in them that getting to the actual credit can only be achieved by narrowing down the list to actual films and getting verified credits for them. In the case of Nunez, there are only 34 actual films for those 193 "titles" and of those, 0 have been verified to match the CLT result "winner" (107/228) which just happens to be the fully accented, comma-less, IMDB version of his name. Note, there are still 11 of those 34 titles not yet verified.
The most important thing is to get as many profiles corrected to match on-screen credits as possible, preserving linking, even if the CLT common name turns out to be the wrong choice, where you know there are variants or it's pointed out by the voters. The more profiles that get corrected credited as data, the more the CLT will start to give more accurate results.
I've not been voting no to common name choices that look to me like it will probably end up changing down the road, because I don't have the time to research all the variants. (I don't necessarily vote yes to them either) I spent over 5 hours creating and editing the posts on Nunez alone, looking at all 371 CLT profile results line by line making a spreadsheet trying to count title/profile totals, and PMing several users asking for help. Most of the people I asked were pretty quick to give me results, and every one of them got a green arrow for their time. I think the CLT needs a little work, to make it more efficient to use, like sorting by original title, sort by role as another viewing option, (which might help group the films) and it would be nice to have a count of original titles.