Interesting but I'm not surprised. Over twenty years, people retire at which time there are no standard mechanisms I am aware of by which a typical academic institution takes responsibility for archiving that data. And, I have stacks of data on zip drives that I and most others no longer have means of accessing. My 3.5" and 5.25" floppy disks have long ago been discarded.
I do have all my old notebooks predating the digital conversion suggesting that the written word may be less ephemeral than the digital word, at least for me. I did have a huge box with all my old films, stored in a room in the department. But that was deemed discardable by some anonymous person a few years back. So, boxes of hard, raw data are subject to loss as well. It is worrisome that over the last twenty years, pretty much all of my raw data has been stored digitally and likely will just disappear.
Morevover, much of the raw data, collected over many years by different individuals, would be hard for others to decipher. We tend to accrue data with specific projects in mind, after which multiple independent studies are immediately collated into graphs and tables, and then use the publication as our primary method for archiving substantiated results. Raw data archives are missing the 'why were experimental directions pursued and others abandoned', which only those of us who were there can state.
Published data represent only a fraction of our output over the years. I have large volumes of equally substantiated but unpublished data collecting dust. That data typically consists of piecemeal investigatations into areas never followed up in the lab, often because someone left the lab before 'completing' the study. I know of many intriguing findings within that data waiting for someone else to run with it. My biggest lament is that no one will ever do it because current publication practices depend on us reporting the 'complete' study rather than reporting 'Hey, what an interesting finding. Who wants to finish that?'
In short, I worry more about continuity of substantiated work found by us to be interesting but no longer pursued. I wish there were some way, at the end of my career (which sadly appears to be now), of presenting some outstanding findings for others to puruse. Some of it is decades old and subject to the criticism (beyond being incomplete) that the methods are outdated and must be followed up with today's capabilities in order to be published. Exactly why I wish there were some way for these short reports to see the light of day. So yes, raw data archives are important from the context of preserving the historical record. But archives in boxes for historians to look at a century from now is only part of the issue. There are publication roadblocks that impede us from letting others know of the nuggets of ground-breaking science buried within those archives.