Great sculpture finds @ChrisAldrich and clever how he attempted to customise preview per site, but hasn’t og:image tags basically solved that particular ‘problem’? — #openweb#techhttps://t.co/2NsNgJ8KrL
As Francis Urquhart might say, “You might very well think that…”
The whole point of that post is to show that og hasn’t solved it. There are too many flavors of metacrap and no standards. And worse og is not only not “open” it’s a DRY violation.
It’s metacrap–I know, I know–but I’ve been thinking about easy ways to use Open Graph Protocol meta data to add contextual Twitter cards to some of my content when syndicating posts to Twitter. My goal is to leverage the speed and ease-of-use of Micropub clients while doing as little as possible manually, but achieving as much parity between posts on multiple sites.
I’m particularly keen to do this to syndicate/share more of the articles I read and post about on my site without adding additional manual work on my own part.
Outline of Some Resources
The Post Kinds plugin for WordPress parses URLs for me and pulls in data to create reply contexts for a variety of posts like bookmarks, reads, watches, listens, etc. Since Post Kinds doesn’t display featured images (yet), I’ve also been using the External Featured Image plugin to display the featured images from the original to add to the reply context of my posts as well.
In addition to all these I’ve been using the All in One SEO plugin to easily add an SEO layer to posts without having to do much thinking about it. AIOSEO recently upgraded their UI and features in the last year, and yesterday I upgraded to the newest v4.0+. One of the new features it’s got is the ability to add default fields or pull in pre-existing custom fields to output OGP meta data.
Start of a Solution
So I got the idea that since Post Kinds and External Featured Image plugins are pulling in and displaying the sort of data I’d like to show in Twitter cards, I figured why not use them? While metacrap is a DRY violation, the fact that it’s automated for me and is based on data I’m actually showing visually on my website makes it feel much less dirty. It also has the benefit that it helps make some of my syndicated content look and feel on Twitter, more like it does on my website. This is also a problem since Twitter hampers how much data I can syndicate in a single post.
I’ve still got some issues about how to deal with the Post Kinds data, but after a bit of digging around, I discovered the image URL for External Featured Image plugin is hiding in the _dcms_eufi_img field. So I can make the default Twitter settings in AIOSEO pull the external image by setting Default Post Image Source to Image from Custom Field and set the Post Custom Field Name to _dcms_eufi_img.
Since a lot of my posts are reads, bookmarks, etc., this works well, but I can easily override the settings for articles or other custom posts which I make less frequently.
Hopefully I can figure out the settings for Post Kinds to get the rest of the default fields to map across. I’m happy to hear ideas on what field names I’d need to use to get the Post Kinds Name and Summary/Quote fields to map over for the og:title and og:description respectively. Ideally I can manage to get it done without needing to get a subscription to the pro version of AIOSEO which also has support for custom taxonomies which is how Post Kinds works.
Since my theme has relatively solid microformats support, and I have plugin infrastructure to allow easy syndication from my website to Twitter through micropub clients, this last bit for creating Twitter reply contexts helps close some of the loop for me in my syndication workflow while keeping as much context across platforms.
Example
Here’s a visual example of a native post on my site and the corresponding syndicated copy on Twitter. There are some differences, mostly because I don’t have as much control of the appearance on Twitter as I do on my own site, but they’re about as close as I can get them with minimal work.
I wasn’t seeing it before, but after upgrading to All in One SEO to 4.0.9 from 3.7ish, the Twitter plugin was throwing in some Twitter card metacrap that was causing a conflict. I’ve turned off the Twitter plugin which shouldn’t affect much since I wasn’t really using much of it’s additional functionality.
Metadata is "data about data" -- information like keywords, page-length, title, word-count, abstract, location, SKU, ISBN, and so on. Explicit, human-generated metadata has enjoyed recent trendiness, especially in the world of XML. A typical scenario goes like this: a number of suppliers get together and agree on a metadata standard -- a Document Type Definition or scheme -- for a given subject area, say washing machines. They agree to a common vocabulary for describing washing machines: size, capacity, energy consumption, water consumption, price. They create machine-readable databases of their inventory, which are available in whole or part to search agents and other databases, so that a consumer can enter the parameters of the washing machine he's seeking and query multiple sites simultaneously for an exhaustive list of the available washing machines that meet his criteria.
Doctorow apparently drew out the roadmap for Twitter, democracy, and the future of society in 2001.
If everyone would subscribe to such a system and create good metadata for the purposes of describing their goods, services and information, it would be a trivial matter to search the Internet for highly qualified, context-sensitive results: a fan could find all the downloadable music in a given genre, a manufacturer could efficiently discover suppliers, travelers could easily choose a hotel room for an upcoming trip.
A world of exhaustive, reliable metadata would be a utopia. It’s also a pipe-dream, founded on self-delusion, nerd hubris and hysterically inflated market opportunities. ❧
Apparently this also now applies to politics and democracy too.
Annotated on August 06, 2020 at 09:12AM
When poisoning the well confers benefits to the poisoners, the meta-waters get awfully toxic in short order. ❧
If we look at Twitter as a worldwide annotation tool which is generating metadata on a much tinier subset of primary documents (some of which are not truthful themselves), this seems to bear out in that setting as well.
ref: Kalir & Garcia in Annotation
Annotated on August 06, 2020 at 09:17AM
This section highlights why relying on algorithmic feeds in social media platforms like Facebook and Twitter can be toxic. Your feed is full of what they think you’ll like and click on instead of giving you the choice.
Annotated on August 06, 2020 at 09:23AM
It’s wishful thinking to believe that a group of people competing to advance their agendas will be universally pleased with any hierarchy of knowledge. The best that we can hope for is a detente in which everyone is equally miserable. ❧
The fate of true democracies.
Annotated on August 06, 2020 at 09:33AM