The term thunk is a portmanteau neologism, combining THing and UNKnown..
As a word, it’s nounable – “this oughtta be a thunk” – and verbable – “you should thunk that”.
The homonymic collision with either the jokey future-perfect tense of think (as in, “who’d have thunk it?”), or the computing term, don’t hurt, and possibly help.
(Had a related domain name been available, this project would have been called “unkunk" – at least as a code name, for a while. )
Other than the words-only context and text, previously described, what makes up a thunk?
Here’s an expanded view of the last post’s example thunk, as it would appear after toggling the ‘links’ footer control. These extra details displayed only when requested are called the thunk’s annex:
Elinor Ostrom, American political economist AElinor “Lin” Ostrom (born Elinor Claire Awan; August 7, 1933 – June 12, 2012) was an American political economist. She was awarded the 2009 Nobel Prize in Economics, shared with Oliver E. Williamson, for “her analysis of economic governance, especially the commons”. links « Nobel laureates, Nobelists, female economists, tragedy of the commons /t8rA62d/elinor-ostrom
The list of short phrases under the text are the hooks, alternate words/tags which should find the thunk, even if they don’t appear in the text. (Think: keyword-stuffing, for good.)
Although the main text allows no blue hyperlinks, up to three supporting links, with title and URL, may appear in the annex.
At bottom comes the thunk’s permalink, made from its compact unique ID and its slug. The autogenerated IDs begin with a ‘t’, then continues with case-sensitive alphanumerics, for example: ‘t8rA62d’. The slug, editable by contributors, helps to make the URL more self-describing.
The last editor and time-of-edit are of course tracked for each revision, but like Wikipedia, individual authorship is not emphasized: it’s just available for understanding edit histories.
Finally, what’s that letter ‘A’ at the top left? That’s this thunk’s grade, a single-letter quality indicator that’s self-reported by editors. Only ‘A’ and ‘B’ graded thunks have full search visibility; ‘C’ and below are essentially drafts, with limited visibility until their content, style, or supporting links are improved.
In Thunkpedia, what is a ‘thunk’?
A thunk is the smallest useful chunk of knowledge, offered as an increment from what you already know… to what you’re trying to find out.
In the system, it’s a capped-size bit of reference text, no more than a few sentences, with some contextual metadata that helps readers, authors, and indexing systems. The default minimal display of a thunk looks roughly like this:
Elinor Ostrom, American political economist Elinor “Lin” Ostrom (born Elinor Claire Awan; August 7, 1933 – June 12, 2012) was an American political economist. She was awarded the 2009 Nobel Prize in Economics, shared with Oliver E. Williamson, for “her analysis of economic governance, especially the commons”. links »
That top line, in fainter type, is the thunk context: a leading one-line label that helps with the interpretation of the main text. There’s no requirement for it to be unique – so don’t think of it as a ‘title’.
The main body, in normal-weight type, is the thunk text: concise reference knowledge, on a focused topic, in just one or a few complete sentences.
In a list, thunks are reminiscent of search engine results with ‘snippets’: easily skimmed. But unlike auto-generated excerpts, each thunk is written to be understandable whether standing alone, or in a mixed list.
Notably, neither the context nor text contain outbound blue hyperlinks. A thunk is not, primarily, for sending you elsewhere but instead for answering your need inline.
The small ‘links’ footer will toggle the display of extra info – but more about that next time!
There’s been slower progress than I’d hoped, but I’ve recently been improving the search and contribution interfaces.
I’ll open the prototype for wider use as soon as there’s some support for user-to-user review and interaction, and a fresh import of Wikipedia abstracts as starter data. Let me know how to reach you if you’d like to get more involved.
Here’s the presentation I did last Friday at the Books-in-Browsers conference.
I’ve added a few screenshots to the slides below to simulate the brief demo given. As you’ll see on slide 25, the name of the site at launch will be ‘Thunkpedia’, as the central organizing principle will be small ‘thunks’ of standalone reference information.
(Updated 11/2 with video.)
I’ll be speaking about Infinithree at the 2011 Books in Browsers meeting (#BiB11) in late October, put on by the Internet Archive and O’Reilly Media.
The title of my talk is, ‘Beyond the Encyclopedia: non-linear reference works’. Astute readers will find fresh details about project plans in the talk abstract:
Should an online reference work aspire to the ‘encyclopedic’ standards of printed volumes? How should reference text change when modern search is assumed as the primary interface? If opening the scope of a reference work to everything ‘true and useful’, what other constraints on contribution and exploration might prove necessary? Can a reference work that seeks to offer canonical information on many topics use the techniques of social news and social question-and-answer sites to encourage casual, incremental contributions?
What if Wikipedia before MediaWiki had started on TiddlyWiki, rather than UseModWiki?
The Infinithree Project (working title) is exploring these questions in the context of a new, Creative Commons-licensed reference work. Prelaunch discussion is occurring at http://infinithree.org. By October, public launch under a different name is likely, so all titles and themes subject to change after contact with the authordience.
(Welcome to all the new followers arriving after this HN thread!)
Emilio Velis (@dubsnipe) asks via the Tumblr question function:
As far as I understand, Infinithree aims to be a non-rigorous source of information, instead of a source full of data that has to be validated by a crowd. How about if the project aims to be some sort of an add-on or sister page based on Wikipedia URLs? For example, there is some useful information that can’t be on Wikipedia because it hasn’t been verified. It can grow itself to adapt to already existing Wikipedia articles, and hold more information (as long as it doesn’t turn into a trivia-only site).
Capturing info that Wikipedia can’t (yet) find a place for is definitely a goal of Infinithree. So with respect to Wikipedia’s exact rules, yes, Infinithree will be less rigorous.
However, we still want all info to be true and useful. That implies enforced standards - rigorous in new ways. I think of it like this:
'Wikipedia is an Encyclopedia.' (AKA 'WP:ENC' or '1st of the Five Pillars’.) That’s the banner under which Wikipedia launched, and the rallying principle Wikipedia has used to meet challenges along the way. (And if ‘encyclopedic’ is Wikipedia’s battle flag, notability/verifiability has been its shield, and deletionism its sword.) So Wikipedia has the ‘encyclopedic’ circle above covered.
But Infinithree has the benefit of a 10-year tail-start on Wikipedia. (‘Tail-start’ being the opposite of ‘head-start’, of course.) With a later understanding of mass collaboration, the primacy of search, and variations on the wiki model, maybe we can aim for new platonic ideals beyond the ‘encyclopedia’. Maybe new policies and constraints can yield reliability with less friction.
Infinithree will explore how wiki-nature and other ascendant mechanisms for online creation and curation can cover all of the ‘true’ and ‘useful’ circles above — especially the non-‘encyclopedic’ parts.
With that broad goal in mind, an organization based on ‘sister pages’ would be confining. So Infinithree will emphasize excellent search and contextual navigation, rather than any 1:1 mapping to Wikipedia articles.
Via the Tumblr suggest-a-post system, Inclusionniste writes:
Start with a multilingual version
I’m very interested about a “inclusionist” “encyclopedia”, but I would like to contribute in French, my native language.
I would love to follow the Wikipedia model, with focused sites for every language where there can be an editing community.
Making the core software accept contributions in many languages is relatively easy with modern frameworks and Unicode. Adapting the interface and core explanatory text is also very achievable, requiring a little more expertise and sensitivity.
But the deciding factor for Infinithree’s success in any language is the editor community: how it starts, how it grows, how it evolves.
I can directly collaborate with others in English, to help establish early patterns of quality writing and civil discussion. (I apologize for only being fluent in my native English!)
For every other language, help will be needed.
Perhaps each language can launch when a critical mass of contributors, speaking that language and understanding the project, declare their interest.
Some time in February, we’ll take advance registrations of interest, so people can be notified as the site opens. Based on your suggestion, I will encourage registrants to also declare what languages they’d like to use.
So please, let your fellow francophones (and speakers of every language) know about the Infinithree plan, and follow this blog and
@infinithree@thunkpedia for discussion and updates.
At Quora, I asked for suggestions about how an avowedly-inclusionist sibling of Wikipedia should work. Marius Kempe has offered a bunch of thoughtful ideas, including the recommendation that rather than any license, Infinithree consider putting all material into the public domain:
- Put the content in the public domain using CC0, or at least use CC-BY, rather than licensing it under the restrictive CC-BY-SA license - the SA clause makes it a hassle to reuse the content, as you can only add other content that is under the same license to it: the upshot of this is that you can’t even legally combine public domain material with SA licenses, as you cannot claim copyright to it! I am personally partial to CC0 over CC-BY for reasons I’ve described in , but either is better than CC-BY-SA.
I’m a fan of explicitly assigning material to the public domain – but bidirectional sharing with Wikipedia, and other sibling ‘Creative Commons Attribution-ShareAlike (‘CC-BY-SA’) licensed projects, will be important for Infinithree. License compatibility allows material deleted from Wikipedia to be transplanted easily, and material initially developed at Infinithree to be migrated to Wikipedia when appropriate. Those reasons alone make CC-BY-SA the presumptive choice.
Also, my sense is that many free culture volunteers are most comfortable contributing to a project with the ‘attribution’ and ‘share-alike’ conditions in place. Their fear is that otherwise, proprietary/parasitic forks can dilute the project’s prominence while sending neither people nor improved material back. I tend not to worry about such things – thinking people will know and appreciate the true origin – but see this concern often in copyleft-related discussions.
(Separately, I don’t understand the suggestion that public-domain content can’t be mixed with copyleft/share-alike licensed material. Truly public-domain content is the ‘universal donor’ for authorship, allowing combination into derivative works of any license – isn’t it?)
Much needs to be worked out before the Infinithree site is open for contributions. Until then, this blog is to discuss possibilities and find collaborators – and you’re already helping with your question and well-wishes. Thanks!
My current inclination is that it should be possible to bring over Wikipedia content – but such content would only rarely be imported automatically or in bulk. The point is not duplicating Wikipedia but trying new models – more than just extra MediaWiki installs – to achieve broader and deeper coverage.
Infinithree will operate in the same social/legal environment as Wikipedia, and will need to exercise much the same care with regard to the rights of living people. Still, there should be room for innovation. “True, useful, and civil” is a potential standard looser than “notable and sourced from a traditional publication”
Did you have a particularly ‘Kafkaesque’ policy in mind? :)