The elf Ahlert came back to the wizard Uku with tears in her eyes.
– “What’s wrong?” inquired the wizard and studied the crying elf.
– “It was there. I know it was there, but I couldn’t find it”, sobbed Ahlert. “What?” asked the wizard. “What was where? I don’t understand.”
– “I had a perfect clue for a data chest. I was so keen on finding it. But all I found were some empty chests with tiny chunks of gold and a kind of mixture lying about. But there was absolutely nothing that looked like data”, wondered Ahlert, shaking her head. “So close, yet so far. It was mentioned in a book … ”.
– “Terribly frustrating!” exclaimed the wizard. “Did you look for clues out there?”
– “Did I look for clues?” Ahlert replied indignantly. “I have spent hours looking around, even trying to talk to people from Ogohu Isle where the chest was supposed to be. But nothing. Nothing!”
– “You have to get back in the saddle,” ordered the wizard.
– “I have a horse?” Ahlert asked bewildered. “Now I’m really confused.”
– “Let me explain something to you,” the wizard said to Ahlert, placing a hand on her shoulder and accompanying her out of the basement. “Horses aren’t always in flesh and blood …”
There are plenty of good reasons why data disappear from repositories and similar places. They may be withdrawn due to cost of having them online, if no one is left to maintain access to the data, or for other reasons. However, if the metadata about the data set disappear as well, it will leave humans and machines in an “unfulfilled” state, when they try to retrieve the data, e.g. when resolving a persistent identifier, or following a link to the entry of a repository. Therefore, make sure to leave metadata telling that “yes, the data were here, but they are no longer available”. Leaving metadata may also offer information on the context, the authors and the institution, where the data were created, for those looking for further details.