Should we really need to have type with different structure?
I believe we arriving to the end of the "old school citation" and we need a different approach where "type" is just one metadata like the others fields.
In my opinion, every document should have access to all the field. For example, a "report" should have an ISBN field, ISSN field and DOI.
For example, the journal become very blurry with preprint and repository services (SSRN, arxiv, …) . Those service provides sometimes DOI but are not peer-reviewed. Even some publication does not peer review. What I want is a radio-button "peer-reviewed" (yes, no, unknown).
Also type is very mixed with "medium". I have a hard time for example to add metadata for presentation or conferences that are actually video (YouTube). I don't need a producer for that.
In my opinion, it can be simpler and more powerful to have one structure to rule them all.
In my opinion, every document should have access to all the field. For example, a "report" should have an ISBN field, ISSN field and DOI.
For example, the journal become very blurry with preprint and repository services (SSRN, arxiv, …) . Those service provides sometimes DOI but are not peer-reviewed. Even some publication does not peer review. What I want is a radio-button "peer-reviewed" (yes, no, unknown).
Also type is very mixed with "medium". I have a hard time for example to add metadata for presentation or conferences that are actually video (YouTube). I don't need a producer for that.
In my opinion, it can be simpler and more powerful to have one structure to rule them all.
2. Having additional fields available (like DOI) is planned for a future release. For now, you can manually add any of them in the "Extra" field as needed, for any item of any type. For example, type:
DOI: 10.xxxx
3. I personally like having different types so I can glance through my thousands of entries and know when I'm clicking on a journal article, book chapter, etc. Similarly, for many non-expert users, having the recommended fields associated with a particular item type seems useful. For example, having "publisher" for book but not for journal article is useful in that sense. There could be unusual times when you'd want to add a publisher for a journal article (maybe a very obscure or generically named journal), but (1) most citation styles will ignore it anyway, and (2) I'd say it's better for most users to have it somewhat clear what fields are expected for a certain type, rather than either missing some important information or feeling obligated to add in every last bit of information for every item even when you really don't need it.
4. It sounds to me like you might need to think again about how types work in Zotero. There's a bit of a learning curve to some of them, but in general they're useful and cover almost all situations. Personally I thought that "Conference paper" was the right type to use for a "paper" (presentation) presented at a conference. But it turns out that should instead be entered under the type "Presentation", while a "Conference paper" is for a published paper in conference proceedings. Your example of a video of a presentation is actually one I'd recommend as a Presentation type, with a link to the video on Youtube: it is not actually a video conceptually, but you are just using that recording of it to cite it. (If it was a narrated video with added introduction, or something like that, the situation would be different.) As for medium, it's important to not mix that up with type, because an ebook is still a book, just in a different format (sometimes annoyingly without pages), but the same content.
I agree with you that for unusual items that borders can blur a bit, but for now you can use Extra to add that information (check that the citation styles you're using actually support it though-- DOI is generally supported for all item types, for example, but maybe not publisher, and of course ISBN is rarely printed in any citations anyway), and there are plans to make the fields more flexible for different item types in the future.
It seems to me that there's also a design question for Zotero: is it primarily a citation-generating tool, or is it primarily an information-organizing tool? (Or maybe both.) Some of the points you brought up fit more with the second use (adding ISBNs even if they're never cited, or adding whether it's peer-reviewed or not, again never actually citing that information in a bibliography). The ideas are reasonable, and others have also mentioned a desire to use Zotero to organize more than to produce bibliographies, but that's a different (or at least additional) philosophical direction, so something that the designers would have to think about (not me). Personally I like Zotero being optimized for generating bibliographies, but I do of course also use it to organize, so having some additional features like that, as long as they didn't get in the way of the primary purpose (in my mind) would be fine.
@zuphilip I needed "conference name" for video recording
For the interface you can display only some fields by default, but let edit all of them if needed:
-> "basic fields" that are available to all item type.
-> "specific fields" that are shown when you select a specific item type
-> "show all field" button, that let you see all fields
@adamsmith I see, this can definitely be problematic. I taught citation style would not break but simply don't show the extra field. So I guess we have wait 20 years or more has I'm pretty sure journal will take ages to update their style.
I don't know that making the distinction between item types less obvious would be that helpful, though, and it might be more confusing. Renaming a few item types could help ("Conference paper" > "Proceedings paper" would make sense to me for example), but I think many users do actually rely on the hint of which fields are available when they try out different item types, and just go with what seems like it works. So giving more options could help in some ways and hurt in others.
I used the database to run some experiments in the past. For example scrapping all blog post for a particular collection to check the content and do some automatic text analysis. I can not do that if the structure is not 100% right. Using "extra" is not an option neither as it will add a lot of complexity in programming such script.
The current state is not future prof and the "extra field" is an hack that is going to be hard to migrate to a new structure.