[fitsbits] Preservation of digital data guaranteed to last at least 500 years

[Jessica Mink]

Well, we have data from up to 130 years ago in 500,000 photographic 
plates in the Harvard Plate Collection, of which over 200,000 have been 
digitized so far, into FITS image files and source catalogs on spinning 
disks stored in multiple locations. The plates are more permanent that 
the disks, unless they are flooded by a water main break...

I'm currently working with digital spectra from various Mt. Hopkins 
spectrographs which go back as far as 1979. They've been migrated across 
various disks and operating systems over the years but are still 
basically intact, though as we migrate the archive into a cloud service, 
we are finding that the data is not quite as uniform as we thought that 
it was. I've been involved for the past 30 years, though, so I mostly 
understand the nature of the variations, and the PI of the original 
project is still around, though busy with Kepler and Tess...

-Jessica

On 04/20/2018 04:34 PM, Rob Seaman wrote:
> Astronomical data retain significant time domain value indefinitely,
> e.g., eclipse data constraining Earth rotation or historical accounts of
> novae / supernovae being associated with current day SNe remnants.
>
> Long term archives require permanent institutional oversight and few
> institutions other than the Vatican are safe bets to exist in 500 years.
> Maybe some governments, but in a world being run like a game of Risk it
> isn't obvious which will prevail. Some universities, perhaps, but which
> observatories? As journals go digital have they become more or less
> permanent?
>
> The Long Now Foundation (http://longnow.org) exists to address such
> questions, as well as create durable artifacts like the 10,000 Year
> Clock. We might seek such a partner if the usual funding channels don't
> seem to address the requirements. This would require long term buy-in
> from some professional organization such as IAU or AAS.
>
> Rob
>
> --
>
>
> On 4/20/18 6:53 AM, Tom McGlynn (NASA/GSFC Code 660.1) wrote:
>> Very interesting to hear about!
>>
>> William Pence wrote:
>>> ...
>>
>>>   These islands are a protected demilitarized territory and the
>>> archive is located within a mountain that has been proven to resist
>>> nuclear disaster.
>> Curious how they feel that this has been proven?  Unless there have
>> been some nuclear disasters kept secret!
>>
>>> Can any astronomical observatory claim to have anything close to this
>>> level of data preservation security??
>>>
>>> -Bill
>>>
>>
>> I would doubt it, but it seems to me that this begs the question of
>> what is the appropriate level of security for our data. I'm not sure
>> that the Vatican's approach would necessarily be appropriate for
>> astronomy data. While our observations may have historical interest in
>> 500 years -- e.g., the plates that Hubble used to first suggest that
>> the universe is expanding -- and thus valuable that way, they are
>> unlikely to be have very much scientific value (in the sense of
>> needing to revisit the original datasets).  And in an age where our
>> observations are increasingly digital,  it is the replication of
>> critical datasets across many institutions that is probably their
>> greatest assurance of survival.  Nor would I think that the threat of
>> physical destruction is the real threat to astronomical data. That
>> seems more likely to be difficulty maintaining the associations with
>> calibrations and tools used.
>>
>> It's a really interesting question since I think that our archives are
>> currently focused on preserving data generation by generation. We
>> don't currently build archives such that the data will be available in
>> the future regardless of whether the archive continues to maintain
>> them.  Maybe it's all just an attempt to keep job security!
>>
>>     Tom McGlynn
>> passing on the responsibility to the next generation to continue to do
>> so, rather than
>>
>> _______________________________________________
>> fitsbits mailing list
>> fitsbits at listmgr.nrao.edu
>> https://listmgr.nrao.edu/mailman/listinfo/fitsbits
>>
>
> _______________________________________________
> fitsbits mailing list
> fitsbits at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/fitsbits
>