Page MenuHomePhabricator

Improve the editing reason labels and descriptions to be more precise
Open, Needs TriagePublic

Description

Problem: As a user, when I'm editing Wikidata from Wikipedia with Wikidata Bridge, I don't find an applicable option for my edit, for example when:

  • I'm completing or making more precise the original value, which I already considered correct but less detailed.
  • I know the original value is not applicable today but I cannot determine if it was correct at some point in the past or not.

options.png (556×612 px, 56 KB)

Suggestions (WIP)

  • change order of radio buttons to "i updated" first.
  • change "I updated an outdated value / The previous value used to be correct but now is outdated." to "I updated the value / The previous value may have been correct but now is outdated."
  • change "I corrected an incorrect value / The previous value was not correct and has never been." to "I corrected or completed the value / The previous value was incorrect, incomplete or less precise."

Event Timeline

Hi Abian, thanks for your report :)

The options we present on this screen are connected to the type of edit that the Bridge does on Wikidata. For example, it can directly update the existing value of a statement (option 1, correct), or add a new value (option 2, update). In the future, it will also be able to deal with statements having several values, with ranks, editing the labels of a value having the datatype Item, etc. So behind each option lies the adapted action on Wikidata.

We think that completing a value or making it more precise can be considered as "correcting". The action behind it would be the same (editing the existing value).

About "I know the original value is not applicable today but I cannot determine if it was correct at some point in the past or not.", is that a theoretical example, or did you encounter this situation in the past? If so, can you tell us more about this example?
If someone is facing this situation, what should be the related action on Wikidata?

Overall, we are trying to keep the number of choices short, so the user doesn't feel overwhelmed with options that they didn't know were even possible. We do not offer an "other reason" option, as it would not be connected to any specific action on Wikidata. We acknowledge the fact that they are options that are not perfectly covered by the Bridge, but we hope that the most usual cases are covered.

Hi, Léa; I hope you have enjoyed the summer (which isn't technically over). :-)

Thank you for the explanation. The risk I see in having two such opposite and assertive options ("incorrect", "was not correct and has never been", etc.) is that, in the grey area, or in case of doubt, users might end up choosing randomly, and that, from what you say, might not be ideal because it would have consequences on how values are finally reflected.

We think that completing a value or making it more precise can be considered as "correcting". The action behind it would be the same (editing the existing value).

Okay. Then the description might be incomplete; for example, if I change a territorial administrative unit to a lower one, I wouldn't consider that the previous value "was not correct and has never been". I personally would think that I would have improved the original value, but not that the original value was wrong, and I wouldn't know what to choose. In this situation people might choose arbitrarily (if they didn't have much time) or they might feel forced to ask about the consequences of these decisions and about Wikidata's policies on when to add values and when to replace them.

About "I know the original value is not applicable today but I cannot determine if it was correct at some point in the past or not.", is that a theoretical example, or did you encounter this situation in the past? If so, can you tell us more about this example?
If someone is facing this situation, what should be the related action on Wikidata?

First I thought of population figures for a municipality, the number of employees in a company… mainly of quantities, but I think this would apply to any other data type. If I read that a village has a population of 432 according to the infobox, but I have just checked that today there are 276 inhabitants according to an official source, I know that the value from the official source can be considered correct now, but I don't know if the previous figure was correct at some point in the past or not. It seems to me that this situation will occur often, as normally we don't know the full history of anything and we can't rule out that a value may have been correct an unknown number of years ago.

As to what the related action should be... in my opinion, the value should be overwritten if it had no qualifiers or references, and preserved if it had a reference or a qualifier (in this case, the new value should have a preferred rank, and the original value should be downgraded to normal value if it had been marked as preferred). But this is only my opinion, and I know it's a mess; when in doubt, adding the value might be the lesser of two evils. (?)


I am going to make some suggestions gathering all this together:

  • The order of the options could be changed so that the most specific or less ambiguous option ("I updated") appears first. This might let the user know that the wording of the second option is no longer covering the first case ("I corrected" does not include updating, because that option has already been mentioned).
  • "I updated an outdated value / The previous value used to be correct but now is outdated."
    • → "I updated the (or "a") value / The previous (or "original") value may have been (to cover the doubtful case) correct but now is outdated."
  • "I corrected an incorrect value / The previous value was not correct and has never been."
    • → "I corrected or completed the (or "a") value / The previous (or "original") value was less correct, complete or precise."

These are just my suggestions, feel free to adapt or rule them all out if they aren't useful.

Thank you for your extensive reply!

Thank you for the explanation. The risk I see in having two such opposite and assertive options ("incorrect", "was not correct and has never been", etc.) is that, in the grey area, or in case of doubt, users might end up choosing randomly, and that, from what you say, might not be ideal because it would have consequences on how values are finally reflected.

We see your point here. We carefully considered only having two options, rather than more, because the danger of having multiple options is of course that people wouldn’t know which one fits their case best and would also pick randomly. We’re currently considering adding some sort of explanatory text, to help in the situation of doubt.

Okay. Then the description might be incomplete; for example, if I change a territorial administrative unit to a lower one, I wouldn't consider that the previous value "was not correct and has never been". I personally would think that I would have improved the original value, but not that the original value was wrong, and I wouldn't know what to choose. In this situation people might choose arbitrarily (if they didn't have much time) or they might feel forced to ask about the consequences of these decisions and about Wikidata's policies on when to add values and when to replace them.

I’m not sure i understand the scenario correctly, but i would think, if it was a mistake, then choosing the first option “correcting” would be correct, if the unit has changed then the “outdated” option would be the correct one to pick. Can you clarify?

First I thought of population figures for a municipality, the number of employees in a company… mainly of quantities, but I think this would apply to any other data type. If I read that a village has a population of 432 according to the infobox, but I have just checked that today there are 276 inhabitants according to an official source, I know that the value from the official source can be considered correct now, but I don't know if the previous figure was correct at some point in the past or not. It seems to me that this situation will occur often, as normally we don't know the full history of anything and we can't rule out that a value may have been correct an unknown number of years ago.

As to what the related action should be... in my opinion, the value should be overwritten if it had no qualifiers or references, and preserved if it had a reference or a qualifier (in this case, the new value should have a preferred rank, and the original value should be downgraded to normal value if it had been marked as preferred). But this is only my opinion, and I know it's a mess; when in doubt, adding the value might be the lesser of two evils. (?)

That seems like a reasonable conclusion. Is there something that would stop you from doing exactly that in the bridge? Currently we can not display qualifiers unfortunately, but references can already be viewed. What you said about the lesser evil makes a lot of sense to me. Maybe you’re suggestion of putting the “outdated”-option first, would serve a second purpose. People will more likely select the first option when in doubt.

I am going to make some suggestions gathering all this together:

  • The order of the options could be changed so that the most specific or less ambiguous option ("I updated") appears first. This might let the user know that the wording of the second option is no longer covering the first case ("I corrected" does not include updating, because that option has already been mentioned).
  • "I updated an outdated value / The previous value used to be correct but now is outdated."
    • → "I updated the (or "a") value / The previous (or "original") value may have been (to cover the doubtful case) correct but now is outdated."
  • "I corrected an incorrect value / The previous value was not correct and has never been."
    • → "I corrected or completed the (or "a") value / The previous (or "original") value was less correct, complete or precise."

These are just my suggestions, feel free to adapt or rule them all out if they aren't useful.

I will run these past our technical writer. Thank you for the suggestions! They both make a lot of sense to me!

Hi, Charlie. 🖐

Okay. Then the description might be incomplete; for example, if I change a territorial administrative unit to a lower one, I wouldn't consider that the previous value "was not correct and has never been". I personally would think that I would have improved the original value, but not that the original value was wrong, and I wouldn't know what to choose. In this situation people might choose arbitrarily (if they didn't have much time) or they might feel forced to ask about the consequences of these decisions and about Wikidata's policies on when to add values and when to replace them.

I’m not sure i understand the scenario correctly, but i would think, if it was a mistake, then choosing the first option “correcting” would be correct, if the unit has changed then the “outdated” option would be the correct one to pick. Can you clarify?

I'm thinking of the scenario where the reader doesn't consider the original value to be incorrect or outdated but wishes to reflect a more precise/complete value. For example, they want to change "musem" to "history museum", or "cancer" to "prostate cancer", or "Berlin" to "Friedrichshain". In any case, this will hopefully be resolved with the possible rewording (the addition of "less […] complete or precise").

First I thought of population figures for a municipality, the number of employees in a company… mainly of quantities, but I think this would apply to any other data type. If I read that a village has a population of 432 according to the infobox, but I have just checked that today there are 276 inhabitants according to an official source, I know that the value from the official source can be considered correct now, but I don't know if the previous figure was correct at some point in the past or not. It seems to me that this situation will occur often, as normally we don't know the full history of anything and we can't rule out that a value may have been correct an unknown number of years ago.

As to what the related action should be... in my opinion, the value should be overwritten if it had no qualifiers or references, and preserved if it had a reference or a qualifier (in this case, the new value should have a preferred rank, and the original value should be downgraded to normal value if it had been marked as preferred). But this is only my opinion, and I know it's a mess; when in doubt, adding the value might be the lesser of two evils. (?)

That seems like a reasonable conclusion. Is there something that would stop you from doing exactly that in the bridge? Currently we can not display qualifiers unfortunately, but references can already be viewed. What you said about the lesser evil makes a lot of sense to me. Maybe you’re suggestion of putting the “outdated”-option first, would serve a second purpose. People will more likely select the first option when in doubt.

Great point, I hadn't thought of that. :-)

I am going to make some suggestions gathering all this together:

  • The order of the options could be changed so that the most specific or less ambiguous option ("I updated") appears first. This might let the user know that the wording of the second option is no longer covering the first case ("I corrected" does not include updating, because that option has already been mentioned).
  • "I updated an outdated value / The previous value used to be correct but now is outdated."
    • → "I updated the (or "a") value / The previous (or "original") value may have been (to cover the doubtful case) correct but now is outdated."
  • "I corrected an incorrect value / The previous value was not correct and has never been."
    • → "I corrected or completed the (or "a") value / The previous (or "original") value was less correct, complete or precise."

These are just my suggestions, feel free to adapt or rule them all out if they aren't useful.

I will run these past our technical writer. Thank you for the suggestions! They both make a lot of sense to me!

Perfect; from what we've discussed, I think the rewording and the change of order of the options would be enough to resolve this task. Thank you!

Charlie_WMDE renamed this task from Some types of edits with Wikidata Bridge do not have an applicable selector to Improve the editing reason labels and descriptions to be more precise.Oct 2 2020, 8:59 AM
Charlie_WMDE updated the task description. (Show Details)

hey @abian, i've updated the ticket description a bit and now it's with our technical writer for a last verification. When the changes are approved the ticket will be given to the Wikidata developers to implement. Thank you so much for taking the time to give us feedback on this and your thorough explanations and suggestions. I'm happy to improve the feature wherever possible!

Howdy, I'm the technical writer in question. As is probably clear to everyone involved, properly fine-tuning this message depends on a good understanding of what's happening behind the scenes.

As @Charlie_WMDE informed me, the "correct an incorrect value" discards the previous data and "update outdated data" retains but demotes the existing value. I don't see those facts in the above discussion, and in my view they're the most important thing about this issue, the knowledge of which seems to have motivated the wording "not correct and has never been".

In light of this observation and the foregoing discussion, I propose the following wording:

Please select the type of edit you are making:
I am supplying updated information (archive previous value)
I am correcting an incorrect value (delete previous value)

I also changed to the present tense since the edit is not yet complete. I acknowledge this might be a separate issue that has already been discussed elsewhere, so I'm happy to rewrite in past tense if needed.

Howdy, I'm the technical writer in question.

Hi, technical writer Dan! :-)

In light of this observation and the foregoing discussion, I propose the following wording:

Please select the type of edit you are making:
I am supplying updated information (archive previous value)
I am correcting an incorrect value (delete previous value)

The motivation for this task is that users are clear about and make the right choices in the following two cases:

  1. I'm completing or making more precise the original value, which I already considered correct but less detailed (e.g. museum → history museum; cancer → prostate cancer; Berlin → Friedrichshain).
  2. I know the original value is not applicable today but I cannot determine if it was correct at some point in the past or not (e.g. there was a population figure without references and I want to provide the one I just found on the council's website, which is different).

Case 2 seems to be solved with your new wording, which no longer states that the previous value was necessarily correct at some point: "I am supplying updated information (archive previous value)".

On the contrary, case 1 wouldn't be solved yet, as it's still claimed that the previous value was incorrect ("I am correcting an incorrect value") when it's not considered so ("completing or making more precise the original value, which I already considered correct"). I'm sure that the clarifications in brackets will be positive for experienced users ("archive previous value", "delete previous value"), but unfortunately I don't think they help users who don't know Wikidata's policies, i.e. when to delete/archive, to solve case 1.

unfortunately I don't think they help users who don't know Wikidata's policies, i.e. when to delete/archive, to solve case 1

You might be asking more of these small snippets of text than they can perform; we definitely can't explain or even imply the entire backstory and function of the bridge in two short sentences. Add clarifying links if needed, but my recommendation is to keep the text as brief and crisp as possible. (No wording is going to prevent some bad edits and incorrect choices from happening.)