http://blacksheep.rs/knjizevna-nagrada-dura-jaksic-2

What do you think about research data ‘freshness’ in the context of a research repo?

I asked the question and got a great conversation back. Here’s a quick run down of what people thought.

I’m working on developing a research repository for my work place so I’m pondering all the things about making research data findable and useful. One of the things I’m pondering is data ‘freshness’ — apart from data security/hygiene, how long is it useful to keep research data around? And of course, which type of data is useful to keep and which isn’t? I’m also thinking about good research versus not-good research.

I was curious as to whether my smart Twitter friends were thinking about this too. It turns out, they are! and, one question later, a nice Twitter conversation ensued — right when I went to bed, Australia time. So I thought I’d collect some of it here, for efficiency and posterity. Thanks to everyone who joined the conversation.

I’ve not attributed points, instead, all the relevant tweets are below. I’ve put this together in 30 minutes — progress over perfect. Excuse (and contribute). This is not a point of view, it’s a summary.

Here’s the gist:

1. As with all things, it depends! All kinds of things can affect shelf life. What is the research about? And “does the accuracy and explanatory power maintain or degrade across time”.
2. Research data about context/attitude/behaviour is more long-life than research data about things like interfaces or interactions.
3. Discovery research tends to be more long-life research and is often expensive too, so it makes sense to keep it.
4. It’s only useful to see research about interfaces or interactions to see why a change was made, or why it was not made. It can be useful in retros and can help close the loop with designers.
5. Having good documentation can make sure that research myths can be weeded out if/when they’re formed. “Can you back up that myth with data?”
6. If you have a mature design system and you’re working on the same user needs long term, design research findings don’t become ‘stale’.
7. If a problem that’s emerged in research over time is not fixed, it can be useful to have a backlog of research data to show that this is not a new problem.
8. Bold research headlines backed with lots of stories, examples, context, variations etc. may be a good method for making a lot of data scannable in a research repository.
9. Dating activity around a research entry may be useful: perhaps a best-before date and/or a last-used date? Perhaps knowledge ‘dies’ if there are no new sources or comments added after a certain amount of time?
10. Any research done well is worth preserving.
11. And don’t forget data security. Depending on what you’re storing, you may not have the right to store it forever.