Huw Diprose

Publication Date vs. Inception

Hi. I noticed you replaced "publication date" with "inception" on a number of work items, e.g. Translation State (Q120482895). But as I understand it, for written works, these are generally two different things; for example, Lost in America (Q48229) has an inception date of 1912-1914 (when it was written) and a publication date of 1927 (when it was first published). Most of the time, we only have information on the publication date of a work. Pfadintegral (talk) 05:22, 31 August 2023 (UTC)

Ah thanks a lot for reaching out. I've been a bit on the fence about this myself would love a sense check.

The base element is that I've been using this wikidata project as my guide.

I've liked that it makes an abstraction between the work "translation state" which I take to be a root element under which we can organise all else. Then an edition or translation, and even an exemplar (a unique first edition or sole copy perhaps).

At the base, I've been trying to follow their model. In which I think the work is never formally published so cannot have a publication date (there are currently constrains on the publisher property, but bizzarely not on publication date). 🤔

But then to your point on how to understand inception. I think it's a fair point.

Is inception for a work when the idea occured, when it was written or when it was finished... 🤔

I couldn't find any good talk on this (perhaps we should start a thread).

My take is that a work is different from a draft and an idea.

Works may be written years after the idea and go through many drafts.

But what we know is the work usually as published. It's finished until there is a new edition or translation. On that basis publication date as a moment of inception for the work seemed reasonable to me, especially as we lose an accurate publication date to an edition.

Generally where none exists I've been finding the year a first edition was published and keeping it at that.

Sorry for the essay, but wanted to ensure I explained myself in full.

I'd be keen to discuss more if you disagree with anything above? Huw Diprose (talk) 07:39, 31 August 2023 (UTC)

Well, using "inception" in the sense of "first published" has the problem that it makes it impossible to record on the work level that - to go back to the example I gave - Kafka's Amerika was written in 1914 and first published in 1927. I would agree that "inception" should be used for the time when a written work fully existed, not just as an idea, but as a text, but that is still something other than first publication.

Also, the property "publication date" itself has the description "date or point in time when a work was first published or released", so I think it was always intended for this use even if the wiki project table doesn't list it. A work does not have a publisher, but it does have a publication date in this sense. Pfadintegral (talk) 17:34, 31 August 2023 (UTC)

Fair point.

I might put this to the author's of the wikidata project and see if they have thoughts?

You have convinced me that there's something missing on inception... But I am wondering if a qualifier would be a good solution?

So in your example we'd do:

Work:

- Title: Amerika

- Author: Franz Kafka

- Inception: 1914 (Qualifier written)

- Has Edition or Translation: ...

...

Then we could have

Editions:

- Publication Date: 1927

- Publisher: Blah

- Number of pages: Foo

That way the publication details are grouped on an edition which is published, but the work remains an abstract?

Would that both clarify Inception suitably whilst preserving the work / edition split?

Also top triva on Amerika 😁

I'm gonna have to go read up what caused the decade delay? Huw Diprose (talk) 17:43, 31 August 2023 (UTC)

Most of Kafka's writing was only published after his death in fact. What I don't understand is where the reluctance to use "publication date" on the work level comes from, given it is more clear and precise than using "inception" in the same sense. Using it only on editions does not seem like a great solution to me, because that is much more difficult to query when trying to find out when a work was first published - you'd have to query and compare all different editions and even then you'd have to assume that wikidata has an item for the first edition, which is not necessarily the case. I have no objection to putting the question to the project talk page. Pfadintegral (talk) 05:25, 1 September 2023 (UTC)

yeah i am open to changing my mind here.

Mostly I'm just seeking a way of doing books that fits well within a data model that the book project seems to be using.

I get your point about querying, but I'd hope that's where SPARQL and requiring all editions to be linked would help (more complex, yes but easy enough with SPARQL I think).

In exchange, we we have an abstract work, a node for the idea of "The Trial", with only elements common to every representation.

Then editions or translations are specific manifestations of a work (audiobooks, translations, a 1990 edition with a forward not found in the original).

Finally there's exemplars which I think are more like "this one famous copy in a museum which Kafka wrote his own notes in the margin of".

I agree though, that some way of knowing the 1990 edition has text first published in 1925 (and written earlier) sounds good to me.

I just wanna find out from the book folk what the consensus is on how to do that.

I'm beginning to think both work.

Publication date as first publication for a work perhaps? (Though can a work be a work if never published?. Then I guess it just doesn't have one).

And Inception perhaps with some qualifiers to reflect what moment we're taking as the start (idea, draft, written, published)?

I've started a chat over here:

https://m.wikidata.org/wiki/Wikidata_talk:WikiProject_Books

Assuming this takes a while.

What would you like to do about Translation State in the meantime?

I can reset it to have the publication date again until the discussion has progressed? Huw Diprose (talk) 08:27, 1 September 2023 (UTC)