The Persistence Argument for Running Your Own Server Is Wrong

Went to IndieWebCamp this weekend, just for a little bit, mainly to listen to the keynotes and hang out with Ward Cunningham and Pete Forsyth. I love the work these people are doing, but I wanted to kick back against one myth there I see repeated over and over.

There are a whole bunch of reasons for running your own server in the age of platform capitalism, but the one I hear used the most often is “Well, you know what happens — you put all your stuff on a new service, and then they delete it on you as they go out of business!” This is followed by a list of things from Google Buzz to Bebo to Friendster that have gone away, taking your history with them.

The thing is this is primarily a first adopter problem. If you were a person in the mid-00s that joined every new social media site to see what the next big thing was going to be, the experience of losing your stuff when some San Jose company didn’t make their B round is probably achingly familiar. But it’s honestly not an experience most people have.

Most people join social networks when they are relatively established, and in general while these more established sites may still become zombified, they do not die. As an example, here is a post of mine from 2004, still available on LiveJournal.


Now the truth is I don’t think I have a single thing in my house that I’ve made that dates back to 2004, apart from my kids. Multiple moves and predictable breakage have grown us a whole new set of objects in the house. Likewise, I have been running personal servers since 1996 on which I’ve put blogs, political sites, wikis, photos, and the like, and none of that content has survived for 13 years. It’s all gone, rotted away, lost, hacked, or just left out by the curb.

Yet here on LiveJournal, immortalized for all time, is the fact that I not only listened to but liked Phantom Planet as an adult. It’s embarrasing, but isn’t that always the way? (At least it shows I was listening to Death in Vegas. And honestly, that Swift Boat prediction turned out to be correct in ways I had not expected. My Dad was brown water navy in Vietnam, here I was thinking that people would finally learn that these positions in Vietnam were the most deadly positions there were in that war).

Meanwhile, in that LiveJournal post I link out to my wife’s art site, and what do you get when you click that? Link rot — that site came down years ago. It’s gone, with all the artwork that was on it. It’s not even retrievable on Wayback, because the new site owner has a robots.txt block in place.

Why does this happen? Why are self-run sites so fragile? I can’t speak for everyone, but for me it was usually switched credit cards, missed payments, open source bitrot, forgotten subdomains, damage from hackers. The link above to caulfieldfamily was just a thing where we forgot to renew the domain registration after an email change, I think. As another example, this blog, which I’ve run under different names for a decade now, is still missing a bunch of posts that were wiped out by a hacker in 2011 or so, via an injection exploit that I didn’t hop on patching fast enough. After that experience I moved to, and I’ve taken a lot of crap for that, but you know what? I haven’t lost any posts, and I never get DMs at 9 p.m. saying, “Hey Mike, there’s something weird up with your site…”

There’s more. I wrote a statistics textbook in 2010 on WordPress — lord knows where that is. I ran a federated wiki site back in 2014. I don’t want to delete the posts that people have put up, but over the past couple years I’ve spent almost $500 keeping those 50 or so sites up on Digital Ocean. Eventually I’ll take that server down. The same with Wikity. Monthly and yearly charges add up.

The same is true for the students we graduate. Students that do something in Google or Microsoft platforms are likely to have that material long after they graduate. As an example, I worked with a faculty member in 2010 to have her students document fair trade projects they did in an introductory class on a Google Sites wiki. That’s still up seven years later.


But what if the students had put it on their own server? What is the chance that work would still be available to them and to the world? There’s five different sites up there — I think optimistically we’d be lucky if even one survived. A student just starting to pay off their student debt is quite likely to let that server lapse in the years after college.


This isn’t to say that students shouldn’t own their own space on the web. They should. But I am more sympathetic to Jon Udell’s vision of “hosted lifebits” on a big dumb server than I am to students running server software. And I think the selling points of not relying on things like Google Sites and Docs tend to be more around issues of tracking data, and hackability, not persistence. Around issues of what Amy Collier calls digital sanctuary — how do we minimize the surveillance to which our students are subjected to in the course of getting an education? And around issues of personalization — how can we make sure students and faculty can expand and customize these tools in ways that make sense for their communities?

Let me state again: these are good reasons to pursue non-corporate solutions! And on the whole, pursuing solutions outside the normal corporate offerings continues to be a noble goal. But we should be honest about the why of it, and for everyone but the early adopters, persistence isn’t it.




3 thoughts on “The Persistence Argument for Running Your Own Server Is Wrong

  1. Reblogged this on Carpet Bomberz Inc. and commented:
    Hosted Lifebits is the model I’m trying to follow as much as possible. And I’ve been happy on without ANY of the customization or plug-ins. I see other struggle along trying to keep their servers up to date, patched, getting the Apache and PHP and MySQL configured and adjusted. I could do that, but I choose not to, and cruise along enjoying reading and commenting on others blogs. I think Mike and Jon are onto something here. Gimme a Big ol’ Dumb Server any day.

  2. Foolishly wading in with both agreement and disagreement. I don’t think it’s quite as simple as either or, self hosted vs them hosted. And I may make a similar move I might criticize you on, to extend your experience to all.

    There’s a lot of stuff in my home from before 2004. My parents saved every bit of school work. I have my old matchbox cars, artifacts that were my parents and grandparents, and until I recently game them to my step son, my childhood coin collection. I have this stuff because it’s important to me, and I take care of it.

    All of my web content created at the Maricopa Community Colleges from 1993-2006 was wiped from the server (a few of them a security nightmare of perl scripts writing to editable text files, but the rest was plain old inert HTML / PHP). But it’s not lost, because I saved all my work on a hard drive (and have it redundantly backed up). So I have my own stuff like my first web presentation from 1994 If/when flickr goes down, I won’t cry, because I have multiple copies of every photo I have taken. There is stuff I don’t have backups for, e.g. videos on YouTube.

    The stuff that matters to you, you take care of. Or you accept their loss as not critical.

    The sites you lost, well hackers did some, but the lapsed domains, the ones students did not maintain? Well, to be honest, they were not all that important. That is not an argument for one architecture over another.

    It’s convenience. We so easily gor for the convenient even at the experience of loss of control, data mining. The rise of facebook is a prime example of people opting for the least effort they need to do to communicate. Click buttons. Give data. Click buttons.

    I might call on cherry picked examples. What about your Geocities sites? If not for the Archive team, lost. That was well established. Posterous was pretty established when Twitter yanked the plug.

    But I cannot discount you completely. Student projects probably will love a longer shelf life on blogger and, Dressed with tracking ads, but long lived. Tumblr? What are the long term odds on that?

    Those hosted lifebits were a lot shaking with that DynaDNS attack a few months ago. Or when Amazon’s cloud has its inevitable guffaws.

    I would say anything import solely hosted on your own server or someone else’s is at risk, and if it’s important, if it really really matters, you should make sure you have your own copies stashed somewhere else.

    I’m intrigued by the Big Dumb Servers but do we really have ny examples? The hosted blogs et al are far from dumb. I’m with you conceptually, but it seems that the overhead on what one might have to do for doing the client side work may be too large a convenience mountain to overcome.

    What matters most is that you care enough about whatever is hosted to care for it, not the server. IMHO.

    • Wading in with both agreement and disagreement is never foolish.

      I don’t think we disagree in principle, but as a person who can be disorganized at times, I do think I’ve lost *some* things that were truly meaningful through accidental inattention. The difference between self-hosting and social media shutdowns is kind of about whether you’re better at managing a lot of little bumps in the road (self-hosted) or the very occasional dinosaur-killing meteor (corp-hosted). I tend to do better with meteors for some reason. But mileage, variance, etc.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s