Monday, August 17, 2015 at 6:21 AM

Criteria for future-safety

I've written about a future-safe web many times -- the idea that what we're creating on the web should persist. Will the ideas we publish be there years from now, so others can know who we were, what we did and what we thought?

The concern is that the record we're creating is fragile and ephemeral, so that to historians of the future, the period of innovation where we moved our intellectual presence from physical to electronic media will be a blank spot, with almost none of it persisting.

If, for example, this website were to persist, you would be able to read these words, at their permanent address, many years into the future.


I made a list of some of scenarios that illustrate what I mean by future-safety.

  1. Some sites say you can download a full copy of all your work, but if the format is proprietary that doesn't count for much, and if they make it hard to do, or slow, that takes points off as well. There has to be some easy way to do something with the downloaded content.

  2. The highest-rated system would be one that's hosted in static HTML on the server of a long-lived institution. For example, I think the RSS 2.0 spec is well-situated for longevity. It's hosted on a static server at, along with other static content for the law school. Harvard has been around since 1636, that would seem to bode well for it being around in 2115, one would hope, and perhaps 2215? The Library of Congress, established in 1800, would be an excellent place to put long-lived public hosting.

  3. Dynamic content on my Windows 2003 server running on Rackspace would get a low score for longevity.

  4. Something you publish on Medium would get a low score, since the content is part of the business model of a revenue-free startup that's raised a huge amount of money. Chances that the writing survives long-term, relatively low. And Medium is especially dangerous because people are storing historically significant writing on their servers, with no provisions for longevity.

  5. If a service such as Medium offered a chance to mirror content on another site, that would dramatically improve the rating. If it were automatic, default-on, and the mirroring site was a static site of a long-lived institution, it would get the same rating as the long-lived site.

  6. Another way to achieve longevity would be to add an API that allowed it to be part of an openly implemented web content management system. That way other developers could implement mirroring from the private site to a public, static, long-lived one.

  7. A fantastic case-study is Sourceforge, last generation's GitHub, that's now putting malware in the archives (according to reports). When authors remove repos from their server, they put them back. Open source licenses make that possible.

  8. Services like and would get relatively good ratings because they have extensive APIs allowing them to be part of open content systems. APIs can be used to correct a lot of sins.

  9. No one today would get a perfect score because there's no way to purchase (as far as I know) a service agreement for the indefinite future. That would be a truly future-safe service, if we believe that the vendor is long-lived. (That's a key factor, this is not a service that can, imho, be run by a startup.) The agreement has to include renewing the domain name the content is hosted on.

  10. One more thing, a lot of people say they don't care about future-safety, but there's no reason not to care. Essays are tiny capsules of knowledge compared to video and audio, it is very inexpensive to store writing in ways that it can survive long into the future. If users want this, we will have it. If we had a rating for every service, you could pick and choose based on this feature as well as others that matter: readability, distribution, ease of editing.


On August 24 I posted another requirement for future-safety.

Last built: Wed, Sep 9, 2015 at 1:31 PM

By Dave Winer, Monday, August 17, 2015 at 6:21 AM. Ask not what the Internet can do for you.