Language Order

Carlito: I agree with your suggestion regarding languages without order numbers being listed last.

Your example also shows how IMDb does not display attributes for languages. I don't understand why IMDb accepts the attributes (Valencian) and (some dialogue) without displaying them to users.

@gromit82​ good point. Perhaps a tooltip should show the attribute value.

I just noticed that this also affects search filters, as that movie appears under English (primary) and not Spanish (primary).

The "Search primary language only" checkbox would also benefit from better sorting.

Col: Yes, the language numbers can be added and the problem will go away, but ...

1. We don't know what all the titles with missing language order numbers are or how to find them.
2. Even if we find a title with missing language numbers, we would have to have access to and watch the film to determine the correct order of the other languages. 
3. It may be difficult to calculate what the correct order of the languages is, in terms of how much they are used in the film. (For example, few contributors would be willing to use a stopwatch to determine how much of the dialogue is in one language vs. another, and I don't think IMDb expects contributors to do that.)
4. In some cases, a contributor might watch the film but be unable to assess which language is being spoken at a given time. (Example: Suppose an Indian film was primarily in English but also included Hindi and Punjabi dialogue. I would not be familiar enough with either of those other languages to know when the characters are speaking Hindi vs. when they are speaking Punjabi.)
5. Preferably, the IMDb software should "degrade gracefully" when data is missing by displaying something close to what the contributors probably meant, rather than the opposite of what they meant.

@gromit82​ These are good points, but the real root cause is that it should not be possible to submit languages without an order number in the first place, and unfortunately a longtime ago we opened a bit of a Pandora’s Box here.

First though, point #5 above is especially relevant and I could not agree more, however, with a long and ever updating list of features to add and bugs to fix in the software, we have to focus on those issues which cannot be addressed by fixing the data instead.

The situation we are in now is that there are a large number of titles in IMDb where the language(s) attached to a title are accidentally in the correct order even though they lack order numbers. If we start inferring that missing data implies a certain order number then the situation could get worse in ways which are difficult to manage, and then in turn we would require increasing amounts of logic in the software to work around what is still fundamentally a missing data problem.  

Using the first rule of holes (if you find yourself in one, stop digging :-) we will locate the old “enforce a language order number at submission time” ticket and add a link to this conversation to at least refresh the priority.   Once we have that, it’s a little more practical to consider building a process to detect and fix the obvious issues as a one-off clean-up job where we can review corner cases and get back to complete information (thus closing this Pandora’s Box).

We have learned the hard way over nearly 34 years of IMDb that working around data issues in the front-end might seem simple and attractive, but it is rarely the right thing to do.  I will see your “degrade gracefully” and raise you a “root cause analysis” :-)

Hope this helps. 

If we start inferring that missing data implies a certain order number

Not sure anyone wants that, but in the current situation you are ignoring present data and inferring that entries without order numbers should be listed before those with order numbers.

I am not sure about requiring order numbers. No other data type requires them. As gromit said, it requires knowledge that a contributor may not have. We have to be able to contribute even for projects which have not been released yet.

Hi @Col_Needham​, thanks for changing the post to "problem" instead of "idea".

It should be noted that, in the header row of the Languages table, next to order, there's a link to the documentation for credit order instead of the more specific language rules.

Another link appears next to the row for each language being added or corrected. It shows the Key Submission Rules.

At first glance, the texts look quite similar, however, the text on contribute.imdb.com/updates/edit does not mandate using order numbers, but the text on help.imdb.com/contribution does, which seems inconsistent.

The first version of the rules relies on the order in which each entry was added to determine which is the primary language; the second version requires an explicit order value.

What's not documented is that an empty order field, even in the last entry, even if numbered languages exist, could be prioritized over others, as if the order was zero.

Arguably, that is the root cause of the issue at hand: sorting the existing language data.

In tables in which only unnumbered languages exist, the order of entry must be presumed as the correct order, as documented.

When both exist, do contributors use:

A) empty for secondary languages and numbers for primary; or

B) numbers for secondary languages and empty for primary?

The site already infers, in effect, that the missing data means B.

I accept your arguments for caution and defer to your experience with the data, but it stands to

reason that A is the correct answer, and in cases where B was intended the rules were not followed.

Whether the solution is front-end or back-end a determination must be made as to the inference of what the missing data means by default.

The situation we are in now is that there are a large number of titles in IMDb where the language(s) attached to a title are accidentally in the correct order even though they lack order numbers.

Why would those be considered accidental if they are supposed to be added in order?

Correct me if I'm wrong, but within the same order value, the order of entry is preserved, no?

Do you mean that a "large number of titles" have mixed empty and number values where the correct primary language has an empty order but secondary languages have numbers?

@Peter_pbn​ 

Not sure anyone wants that, but in the current situation you are ignoring present data and inferring that entries without order numbers should be listed before those with order numbers.

This is another lesson we have learned the hard way — it is better to keep things working as they are today and wait until we can fix the root cause vs. trying to patch things with interim solutions.  This is because we understand the issues today, and as we have already noted in this case, any impacted titles can be quickly fixed by editing the data.  If we start messing with the software and changing assumptions without fixing the root cause of the problem, there’s a non-trivial chance that whatever interim change we make, the situation will become worse.

Hope the extra context helps. 

Hello all,

Thank you for your patience. Our team is actively investigating this issue. As soon as there is an update, we will provide an update here.

Hello All,

I can confirm that this issue has now been resolved.