This is a question I’ve been thinking about recently and I’m hoping some of my FriendFeed friends can chip in with some ideas. I’m looking for criteria by which if I were to give you an instance of such an ‘artefact’ you would say “Yes, that’s science” and not just some other thing. It’s important to understand that I don’t care how it’s realised/implemented. When answering think about what it should have as well as do or provide, to whom (or what) and for what reasons. Non-functional requirements are welcomed too. To get you thinking about this question complete the following sentence: Most YouTube videos don’t count as the smallest scientific artefact because … - Dan Hagon

The smallest possible scientific artefact is a statement with sufficient associated context that makes it possible to state whether it supports or falsifies a hypothesis? "We found water on mars" does not qualify. "The mars lander has generated images in which a reflective substance is uncovered and then dissappears, consistent with the expected behaviour of solid water at ambient pressure" does. - Cameron Neylon

See also: http://friendfeed.com/science... - Bill Hooker

Thanks @cameron for getting the ball rolling on this. As this is essentially a requirements analysis exercise I'll hold back from critiquing any response but will instead come back with unresolved issues. For instance, in what ways must the artefact support reuse and is it essential that others have peer-reviewed before you can accept it as true science. What I'm thinking of here is if you could make the artefact as small as possible you might be able to make it small enough that you not only strip out all the clearly unnecessary baggage but you might also be able to dispense with things which are normally essential; i.e. does peer-review apply to easily checked one-liners? - Dan Hagon

@Bill I must admit I noticed Steve's post earlier and then completely forgot about it - the idea must have rattled around my head, only to come out a little later in this form. If the two threads differ I think it's more because I'm approaching this question very generally, from the point of view of a software engineer and purposefully not trying too hard to envision how these ideas could be done practically. - Dan Hagon

I don't think peer review is important in determining whether we are talking about a scientific artefact or not. Otherwise peer review has to be a magical process for making a statement "scientific" which I would be deeply uncomfortable with. Peer review is nonetheless important but it is a process for determining the reliability, relevance, and expanding the context of a given artefact. Would also argue that smallest possible unit is smaller than the smallest publishable unit (or at least the smallest unit that is worth the effort of publishing in the sense of "formal publication") - Cameron Neylon

What about "1 mL of water from [GPS coords] contains 30,000 +/= 500 coliform bacteria" - Mr. Gunn

I'd say yes to that but I think I may need to modify my criterion to include the necessity of a link to "the raw data" i.e. provenance information for the assertion. Or maybe that is a question of the"quality" of the statement rather than its "scientificness" but it seems that it isn't useful without some provenance. - Cameron Neylon

"...as measured by EPA testing technique #872873" Better? - Mr. Gunn

is better but is still an assertion...link to notebook containing further links to GPS output and bacteria measurement...and then on. Each of these is probably a worth scientific artefact and they all add more weight/context to the original assertion. But you can never go the whole way back because there is infinite recursion here. At some point you accept the weight of evidence. I might take your word for it but want to be confident there was more backup info there if needed...someone else might want to check your gps calibration first. - Cameron Neylon

Good question, Dan! I'm not coming up with anything clever to say. But I'm interested in trying to generate a test case for a small publishable unit. Maybe we can come at the question from two directions: you theory (this thread) and me experiment (trying to whip up a figure from existing data). - Steve Koch

Now that I reread my previous comment, I agree on the not clever part. What I meant was, maybe I can do something that approaches smallest publishable unit (IMO) and then see what happens and give us some more context for the issues. Mr. Gunn et al. above are essentially providing experimental test cases as well, and more relevant to your original question (artefact; not necessarily publishable) ... practically speaking, there may not be a lot of drive for achieving minimum publishable unit: you'd still like to include extra, sound information if you have it. But there may be more incentive for the minimum artefact -- because this could be something computers could work with. I think J-C et al.'s solubility stuff may be good place to look for those minimum units? - Steve Koch

From the standpoint of the minimal citable unit of science I think blog posts and lab notebook pages (e.g. on a wiki) work extremely well. For example most of our references on this paper http://www.jove.com/index... are of that type. We've published in several journals and have never had an objection from a reviewer that blog posts or wiki pages are not appropriate. I think the key is that sufficient context is provided in the reference so that it is easy for anyone to understand and that appropriate links to raw data are made to support any assertions. - Jean-Claude Bradley

The Concept Web Alliance (or Barend Mons particularly) promoted recently (SWAT4LS) the idea of one (RDF) triple is the minimum: http://conceptweblog.wordpress.com/ - Egon Willighagen

I'm not convinced about a single triple because you don't (necessarily) have context. "A cat belongs to class reptile" is an assertion/hypothesis but without some context, like in Mr Gunn's example. I think we've had a similar conversation before around this (http://cameronneylon.net/blog...). Perhaps the answer is that an _isolated_ triple is no good but one wired into a larger relevant context is? - Cameron Neylon

Presumably for positivists, any 'science' artifact will also have to reference the theory it seeks to refute - otherwise isn't it just data? Perhaps this is what Cameron means by 'context'. - Neil Ernst

Not necessarily. I'm happy with a data artefact that doesn't reference theory. "We measured X by method Y and got result Z". Reference to a theoretical context is a bonus. I think I'm trying to draw a line between isolated assertions and assertions with enough context to provide a way of assessing the relevance of the assertion to whatever question you might be asking. - Cameron Neylon

This would be more of a constructivist approach then. Others can pick (internally consistent) 'units of science' to assemble, po-mo style, new theories and ways of understanding the world. - Neil Ernst

Apologies for the delay in replying to your suggestions and thanks to everyone who's replied so far - it's really given me lots to think about. I think understanding this question can get quite involved so I've decided to switch to using Google Wave for this discussion: http://bit.ly/4rjXbl Please let me know if this disenfranchises you from the discussion and I'll find a workaround. - Dan Hagon

Steve, I agree we should try to put into practice anything which comes out of the theory side because I think this is one of those circumstances where theory and practice can equally well inform each other. I'm not so sure this question is trying to do anything clever either - if anything I probably demonstrates just how naive it's asker is :) - Dan Hagon