Future Media Anxiety

Recently, my father, a member of the local photo club and long-time hobbyist, sent me a document explaining why digital images were inappropriate for archival storage. Aware of the current issues surrounding extinct media and ancient data formats, I began reading the essay to see what constructive suggestions were offered by the author. Unfortunately, I was disappointed to discover that the document was little more than an uninformed rant against archival digital image storage, and as such, I decided to write a rebuttal.

The original document was in Microsoft Word format, which is ironic given it was arguing about future-safe information storage. I trivially converted the document to HTML using Open Office in about five seconds and I have made it available here. I suggest you read Saving Photos: Regular & Digital by Jim Dainis and then consider my rebuttal below.


The following is a quote from the original essay.

By 2015 or so, the computer you buy probably won't have a drive that can read CDs any more.

CDs are _the_ prolific music and data medium of the last twenty years used by virtually all people in all industrialized societies. They are not some obscure just-past-experimental technology like 9-track computer tape drives from 1970. The volumes of useful data on CDs (and DVDs) is immense and there will continue to be economic value in providing readers for the physical media themselves, at least until the point where it is possible to mass-convert that media to a new prolific form.

Just like you kept your 78RPM turntable around, and your 8-track player around-- except that you probably didn't.

But I'll bet you kept your record player, since you had a lot of records. And since everone else had a lot of records, because it was a prolific media type. The two chosen examples were short-lived oddballs and the comparison to CDs is not justified.

Of course, those 78 RPM records and 8-track tapes didn't have your family pictures on them.

This is the achilles heel of this argument. It is because they don't have your family pictures on them that we haven't bothered to keep these old players around. All of the data available on these media was (and is) easily available in better quality on the new media that replaced them. That is, in fact, why the new media replaced the old: not because of superior technology, but because they offered the same content in better ways.

1) If something DOES go wrong with the image-- you don't know until the next time you look at the image. The CD or memory stick still looks fine.

This is no different than having your photos in an album or slides in a box that you rarely open only to discover years later that they are covered in mildew.

As an aside, it would also seem irresponsible to keep digital images on a memory stick for long term archival purposes. These media are sold as temporary storage.

2) When something goes wrong, it might be minor-- or the whole image might become unreadable. You don't know that until you check, either.

This statement effectively says that either something minor can go wrong or something major can go wrong. This is not only self-evident, but it is fear-mongering because it does not responsibly discuss the probability or circumstances of such damage in relation to the alternatives.

3) What goes wrong might not even BE with the image. It's likely to be a problem with the last computer handy that will let you view the image. . . .

This last bullet is really two separate claims. This first claim has some merit, although it is stated inaccurately. Since we have established that access to the raw media (whose first-order data layout is standardized by ISO 9660 and other international agreements) will be available, we are left with the problem of higher-order data descriptions. It is critical to store data in open, standardized formats so that there are many offerings for reading and writing that format, or so that even if there are no readers (which would be extremely unlikely for prolific openly standardized formats), one can use the knowledge of the data format to build a reader in the future.

Unfortunately, it appears that the difference between the concepts of media format and data format is not known to the author since this essay continuously treats the issues surrounding them as one and the same.

. . . Or the last printer handy that connects to that computer. Parallel ports are vanishing from new computers, and in a few decades, USB ports will vanish too. In future what will happen is: somebody in the world can still view your images. If they are still on the discs. But YOU can't view them any more, so you can't even check to see if the discs are OK.

Having already discredited the claim of the last sentence, consider the implication that your image will be unprintable. The argument appears to be that since current printers require current interfaces, they require current computers to print an image. Even if we take this to be true, it does not explain why, in the future, an image read from a standardized format on a prolific standardized media type should be unprintable on future output devices. Since one can read an image into a future computer, one can use a future printer to print it, a future display to view it, and a future network to send it just as easily as one can use a future storage media to store it. Note that this last capability has already been admitted to in the essay.

What WILL work, more often, is real pictures. Sometimes, people even throw those out, especially if they don't know who the subjects are. That's why I advise pencilling names and dates on the back.

Here, the author fails to remember that he is talking about the future, when people will be more familiar with digital media. Allow me to explain.

It is responsible to highlight the issues surrounding new technologies. Essentially, the issue here is that people may make the same assumptions about the new digital storage technologies as they did about the old paper storage technologies. In some cases, the assumptions don't transfer.

Take for example the assumption made about paper media: "Because I can see a picture now, I will be able to see that picture in the future just as easily." People make assumptions to simplify their lives; it is an essential skill for dealing with the world. Often, as in this case, the assumption depends on an further assumption which goes unstated because it is "obvious". In this case, that extra assumption is "because the format will not change in the future". Clearly, in the realm of digital media, this will not always hold. Once we are aware of this, we can plan for it and deal with it appropriately by using prolific open standardized data formats and media types.

If we do this, we can still make the original assumption: "Because I can see a picture now, I will be able to see that picture in the future just as easily." This time, our extra, unstated assumption is "because the format is a prolific open standard that will be usable in the future". By recognizing the properties of digital storage, we can, as an individual, store our data for the future with ease.

If you got a computer in 1994, it is very likely that most of the programs you bought back then will not run on your current computer.

This is not relevant. We have established that it is not important to use the same computer or the same software to read standardized formats from standardized media.

You may wonder-- "Why do I have to write my son's name on every single picture of him?"

Because those pictures will probably not all stay together, as you get more and more grandchildren, great-grandchildren, etc. When they get separated, after awhile some descendant will get a picture and not recognize it.

Leading up to this point, the author, without realizing it, makes the argument for providing metadata along with the images. Annotating images with descriptions of their contents is certainly a good idea. In paper photographs, you are limited to the use of a pencil.

With digital formats, the filename is the most primitive of metadata fields, and can often provide a brief description of the image. Of course, filenames are easily changed, so a more robust solution is to use the metadata fields provided by the digital image formats themselves. Most modern digital cameras already fill in time of day and exposure settings automatically when recoding an image. Good software tools will allow further annotations.

The tedious manual process of annotating all paper photographs with the desired metadata is clearly a weakness in the argument against digital archiving. This is where the computer excels: in organizing, copying, and modifying data. You can use a computer to annotate any number of images at a time, or sort them instantly by queries on their metadata. Instead of having to divvy up your pictures amongst your children and grandchildren, you (or they) can easily make unlimited copies of the entire collection or any subset.

The argument for doing of this annotation, indexing, archiving, and duplication manually falls flat on its face when put against the power of a computer to manipulate and organize digital information.

Can images on storage media fade? Not exactly, but what actually happens may be worse. They get "static" in them. . . . With a digital format, what will happen with missing magnetic particles is pixelation. The image just breaks apart into pieces.

This is patently false. In fact, it is so completely wrong that these statements alone should be enough to discredit the entire essay. Ignore the fact that the essay has been attempting to dissuade the reader from using CDs, DVDs, and flash memories, none of which use "magnetic particles". Ignore the analog term "static" used inappropriately in a digital context. Ignore the error correction of all modern storage media. Ignore the simple solution of redundant copies. What is truly outrageous is that the author claims that when components of the underlying storage go "missing" (whatever that means), the image simply looses some detail. This shows a complete lack of understanding of digital information in general, let alone the properties of specific media types and image formats.

This will happen without any warning. When pictures fade, it happens gradually. A digital image will be perfect-- until suddenly, it's NOT perfect any longer.

This is an inconsistent argument. If the earlier argument about "static" was correct (it's not), then, from a quality point of view, what would be the difference between a slightly faded print and a slightly "static" digital image? How is a slightly deteriorated print still perfect, but a slightly deteriorated digital image not still perfect? It's not.

The true valid point here is that it is important for failure to be observable so action can be taken to deal with it. This is an observation of systems research as a whole, and it harks back to the earlier point about not opening the photo album meaning that you don't know if your pictures are okay. Redundant copies (which are cheap, small, and easy with digital media, or expensive, bulky and not-so-easy with prints) can mitigate the risk here.

After awhile, if you are lucky, you will be old and feeble. After you are 95, will you still be energetic enough to get that transfer made? If not, will your children and grandchildren do it for you? Maybe. Maybe not.

If the party or parties to which you wish to pass own your images do not have interest in archiving them, then your choice of storage media is moot. Plus, a digital copy by a 95-year-old of the future who is familiar with digital image storage technology since he has been using it all their lives is not a difficult or strenuous task. Lugging photo albums to the photo processing store (presuming they still exist) may be another matter.

In principle, digital data lasts forever.

This appears to be a claim made by the author in order to attribute it to those with which he is debating. I can think of no reasonable basis for this naive claim.

If you store your images on your hard drive it can crash. And, sooner or later, you will be getting a new computer. Over 30 years, you will likely get 10 or more computers. Each time, you must move the images. If you store your images on removable media, you have to keep migrating to new storage media-- photo CD's don't last forever.

Again, redundant storage is critical for all archival purposes. The fact that the author continuously ignores this possibility shows that he is either ignorant of it or has no reasonable counter argument. Redundant storage can be highly automated by software and can even storage images in different media formats.

2) Unless you make a serious effort to organize your digital image files, it will be hard to find a picture later for reprinting. By default, files store in the computer, or on the disc with a long, meaningless number as the file name. Making index prints can help. But only if those prints last. See below.

So the author unfairly presumes that all copies of all print images will be tediously annotated by hand, in pencil, but then declares that all digital images will have meaningless filenames and no metadata. And then he ignores the power of the computer to organize those images.

The followup presumption that any thumbnail index has to be on paper and is thus no good because those index prints will likely not last is ridiculous. After all, the essay argues that we should print things on paper in order to make them last!

It ought to go without saying that you can't depend on anyone else to save your images for you. Some people put all their family pictures on a hosting service called PhotoPoint. It's now out of business, and they have no family pictures. That might happen to ANY hosting service.

I actually know some of the PhotoPoint people, and from what I understood, they never claimed to provide archival services, so yes, it would be foolish to store your photos there for purposes other than sharing them with others.

You don't know WHEN you need to migrate images to new storage media. With ordinary photos, a glance will tell you if the image is fading, and you can do something about replacing it, or changing the storage conditions. With large numbers of stored digital images, it's a lot harder to look at them all and make sure they're all right.

First, it is important to note the implicit assumption made here: Because photos are going to degrade, you will have to make new copies sometimes. This, I take no issue with. Unfortunately, the claim that detecting problems with digital images is harder than with prints is false.

It is clearly easy to view large numbers of digital images with a computer. However, thanks to the error-correction and error-detection technologies offered by all modern storage media (and some storage formats), one doesn't have to personally sit and view every single photos to detect a problem. When the computer notifies you of an error when reading from the media or format, get out one or more of your redundant copies and make some new fresh copies while the data is still perfect. Or, even more responsibly, plan ahead of time to re-copy your redundant backups (cheaply and easily), perhaps even using an updated media type or data format. Because the process is automated, all the work is done by the computer. You only need to decide when you want to back up what images to what archival formats and media.

5) Image degradation of digital images is less predictable. You may get a little static-- or, if the bit that goes away is in the formatting, you can lose the whole image at once. It won't happen often, but could.

Yes, there are different failure modes for digital images. All the usual physical problems that can happen to prints, such as fire, flood, etc., still apply, but when the media degrades, instead of a faded print you may get data loss or data corruption. So it is important to know the lifetime and limitations of your storage media regardless of whether your data is analog or digital, since both live in the physical world. Again, the description of the digital failure mode is naive and inaccurate, but the qualification of the claim with a risk assessment is appreciated.

6) Think about what will happen after you are gone, and some relative is going through your things, deciding what to save. Your digital photo files will look to the uninitiated like a pile of obsolete software.

Except your relative is likely to be more technologically aware than you (if they are a descendant) and you have of course labelled the CDs and placed them in some suitable case, just as you would the traditionally family photo albums.

This is typical of the unbalanced arguments made in this essay where the print archivist is described as having great resources, prescience, and patience, while the digital archivist is portrayed as a boob who barely knows what he is doing, failing even to label the images he has planned to give to his grandchildren.

Many of the claims the essay makes about digital information are false, leaving me wondering about the veracity of its other statements. I find allusions to some of the real issues of digital preservation, but I am not compelled by the proposed solutions that make use of print images. All of the archival techniques that are describes as being the solution when using print media are actually easier and cheaper when applied to digital images. While frightening the reader with the volitility of digital formats and digital media, the author fails to recognize the difference between past rare technologies and current prolific ones. Naive presumptions are made about how and why technologies change as well as about how digital technologies work (and fail to work). Overall, the essay misunderstands digital storage and as a result is afraid to use it, and nothing more.

Brad Spencer
February 7, 2004

Valid XHTML 1.0! Viewable with any browser