Cory Doctorow: The Internet of the Dead

As I write this in September, 2012, Charlie Stross and I just returned from a tour with our novel Rapture of the Nerds, a book about – among other things – technological immortality as achieved through brain uploading. Digital mortality was very much on my mind as we crossed the country. I had begun my trip with a few days in Toronto, attending to a strange and new kind of memorial ritual for a close friend who had died unexpectedly in June.

My friend’s name was Erik ‘‘Possum Man’’ Stewart, and I’d known him for decades, since we were high-school roommates. We were both geeks, but Possum was a true hacker, a talented and creative programmer who pursued numerous projects, including a lifelong effort to create spatial games like Pong and Tetris that ran in four or more dimensions. Like me, he was in his early forties, and he was in fine health, apart from an unsuspected weak blood vessel in his brain that ruptured without warning while he slept. When his housemates found him, his computer was still switched on and logged in, and his e-mail open, along with various windows with to-do lists and other random notes from one part of his busy mind to another.

No one – especially Possum – had anticipated his death. It was one of those sudden, freak disasters that challenges the survivors to improvise a response while still reeling with grief. Possum’s loved ones did him proud, with a series of memorials and a packed funeral attended by friends and family from around the world. His friends created books of memories and online memorials, and after the funeral his housemates invited his friends to drop by his room and spend some time in the place where his life had come to an end.

It was while I sat in Possum’s room that I began to think about his computer. It was a homemade Franken-PC that sat under his desk, its wheezy fan making a racket like an ancient refrigerator. After I’d left Possum’s house and headed back to the airport, I got to thinking about that computer. I strongly suspected that Possum would have copied over all the data of his life – all the e-mails and lists and photos and movies and programs and essays and stories and, well, *everything* – onto each new machine, keeping it all live and handy. After all, hard-drives are cheap – especially if you’re building your own tower PC with lots of full-height drive bays – and their capacity increases exponentially, year on year. It’s been a long time since it made sense to keep your archives in a shoebox full of Zip cartridges or floppy drives. If you buy a PC every couple of years, your new machine will almost certainly have more than twice the hard-drive space of your old one. Keeping your data on your live, spinning platter means that it will get saved every time you do your regular backup (assuming you perform this essential ritual!), and if the drive starts to fail, you’ll know about it right away. It’s not like dragging an old floppy out of a dusty box and praying that it hasn’t succumbed to bitrot since it was put away.

Possum never uploaded his consciousness to a computer, but he approximated such a transfer, one keystroke at a time, year after year, filling those noisy, full-height drives with all his secrets, all his creative outpourings, all his minutiae and mundane trivialities and extraordinary profundities. It’s a transfer we’re all effecting, but Possum got a head start on most of us, kicking off the project in the 1980s. That homely, rackety tower under Possum’s desk was him, in some important sense – in the same sense that my laptop holds a good deal of what it means to be me.

So I wrote to his family from the airport and said, ‘‘Look, I know we don’t know what to do with all this data just yet. But we can’t just throw it away. That’d be like throwing all his papers, all his photos and letters, everything, into the trash.’’ What’s more, if you just stick the PC on a shelf or in a box in the basement until you know what to do with the data, there’s a good chance that the data will be lost. When it comes to computers, storage is fraught with peril. The lubricant on the drives’ bearings dries up and the disks seize. They get flooded out, or damp infiltrates them. They get gnawed by rodents, and insects fill them with droppings.

If natural causes don’t get them, then robbery might. Computers are easily pawnable. I have a dear friend, another writer, who died unexpectedly. Her family put away her laptop while they grieved, thinking they’d return to it later. When they were burgled, it was the first thing that was stolen. There was no backup. My friend’s writings are mostly gone now, for good – all that remains is the fragmentary hardcopy.

I knew I’d be returning to Toronto in the late summer before my tour with Charlie – who, incidentally, has written fiction about how stealing someone else’s computer will someday be tantamount to assuming that person’s identity – and so I offered to dump Possum’s data onto a big hard-drive, then migrate the data onto a cloud server somewhere and pay for ten years’ storage, so the family could figure out what to do with it once they’d had some time to grieve, when they weren’t in the midst of their life’s upheaval. They agreed, and I was essentially appointed to be a digital executor.

Possum used an unfamiliar flavor of GNU/Linux – I’m guessing it was some flavor of Slackware – and his computer was booted into it when I arrived at his room with my cheap terabyte hard-drive. Lucky for me, he didn’t use screen-lock, so after tapping the spacebar, I found myself looking at his desktop. Unlucky for me, Possum typed with a Dvorak keyboard layout – a super-efficient alternative to the normal QWERTY keyboard – so I found myself hunting and packing at his keyboard. Eventually, I was able to attach my hard-drive and copy over his whole user directory, but it seemed like there wasn’t nearly enough data there. I didn’t want to miss anything. Between the unfamiliar operating system and the strange keyboard, I was somewhat at a loss.

I’d come prepared with a bootable USB stick with a copy of Ubuntu, the GNU/Linux flavor I use. I rebooted Possum’s machine and started poking around. The Ubuntu stick booted up with the standard QWERTY keyboard layout, which was a relief, until I realized that I would have to switch from hunt-and-peck to touch-typing, because Possum’s keycaps were still Dvorak, so that none of the key-labels were correct. Lucky for me, Possum didn’t encrypt his hard-drives, so I was able to see and then mount the 20-some partitions he’d divided his computer’s two 500GB hard-drives into.

I was right: the data I’d recovered on my first try was just a small slice of the whole pie. Two days later, I’d copied off all that data. It appears that Possum’s standard way of moving from an old PC to a new one was to simply copy over the whole previous drive, then recopy any working files into the new machine’s home directory, leaving the old system intact, so Possum’s drives were full of dozens of copies of the same data, along with dozens of versions of Linux. There were also drives where he appeared to have been testing out other operating systems, and miscellaneous other bulky detritus of a busy, technologically savvy hacker’s life and experiments.

Now I’ve got the data on a drive, encrypted in a ZIP archive with a good, strong password. I could physically mail the disk to Amazon to put onto their S3 cloud service, but I don’t want to risk the post. Instead, a friend with a very high-speed fiber link is going to upload it for me, leaving the drive intact and in place. Amazon has announced a business-to-business cloud storage service called Glacier that is very cheap relative to S3, but it’s ‘‘cold storage’’ – if you want to retrieve your data, you’ll need to give them several hours’ notice. Assuming that Glacier’s early reports continue to be positive, I expect to transfer all the data there and prepay for 10 years’ storage. I’ll put a reminder in my calendar to get in touch with Possum’s family in nine years or so to let them know that it’s time to start thinking about this data. And of course, I’ll give them all the information they need to retrieve the data earlier should they choose.

Within a decade, though, I expect that this will be a much more solved problem. Right now, going through all the data on Possum’s drive is a massive undertaking, and a potentially vast invasion of privacy. In an ideal world, I’d feed the drive-images to a search tool that would index them, de-duplicate any redundant copies, and then make some educated guesses about the stuff that’s suitable and important for family consumption – copies of e-mails to family members, bill payment information, photos, etc.

Presumably, such a tool could be bootstrapped from existing forensic software, such as the stuff used by customs agents to find suspicious data on searched hard-drives. Ironically, these gross invasions of privacy could help us preserve the privacy of the dead (and the living, where a deceased person’s drive contains information that might embarrass a live person). If customs authorities have image-processing algorithms used to find and surface photos that contain nudity, a memorial search engine could use these same algorithms to exclude such photos from a summary of a person’s data.

But personal data isn’t confined to personal computers. Today, your digital persona is as likely to be spread across online services like Flickr, Google Mail, and Facebook as they are to live on your hard-drive. Some of these services have tools for exporting that information, and there are a few digital memorial companies that will host the data online after it is rescued.

The sector is still pretty unsophisticated. There’s no systematic, easy way to prove to Google or Facebook that one of its users has died, let alone establishing which survivor is legally entitled to access the dead person’s accounts. This isn’t the sort of thing you want to leave for months or years, either, as you’d risk having the data purged or hacked. For real geeks, there’s another layer of services: domain registrars who hold the registration for personal domains, certificate authorities who control the secure certificates protecting the servers, and hosting companies who keep the dead person’s websites alive.

This is a very complex area, but it’s also a potentially lucrative trade. I wouldn’t be surprised if my next emotional, unplanned visit to a funeral home included a gentle word about some smart startup that has established relationships with the major online services, cloud providers, registrars, hosting companies, and certificate authorities, and which would mediate on my behalf to get my loved one’s login credentials from them, provided that the funeral director attests to the status of the deceased and my status as his authorized representative.

Such a service will be a godsend for identity thieves, of course. If a funeral director fails to adequately secure his passwords for the service, or is corrupt, or can be tricked, then anyone’s digital life can be stolen in an eyeblink, and comprehensively. This sort of gaping security hole is par for the course in the real world, though (for example, Citibank UK wouldn’t open my bank account based on a Canadian passport and printed statements from their US branches, but they would trust a laser-printed utility bill I could have knocked off in ten minutes in Photoshop), and the market opportunity is so great that I can’t imagine that such a service will be long in coming. Likely, they’ll just get some security company to stick a bunch of meaningless certification badges all over their website promising that they are impregnable and foolproof.

Something has to be done. As Charlie pointed out to me, by 2050 more than half of the Internet’s users will be dead – that is, of all the accounts ever created by Internet users, more than half will have been created by people who have since died. We don’t have the norms, the laws, the software or the markets to deal with this data. Perhaps the Internet Archive will offer to manage peoples’ digital memorials in exchange for long-term deposits of personal hard-drives, not to be disclosed to the public for 100 years. It would be a billion bids for digital immortality stuck in online deep-freezes like preserved heads awaiting medical breakthroughs.

I’m heading back to Toronto this week to kick off my next tour, for Pirate Cinema, and I’ll be handing off the drive of Possum’s data for upload. It’s a small and uncertain thing I do for my friend’s memory, and I’m not sure what will come of it. Better to preserve the data and defer the decision about its long-term fate than to leave it for the vagaries of time. And I’m giving hard thought to what I’d like done with my drives when I go. The passwords for my encrypted disks are on deposit with lawyers I trust, but having tried to make sense of someone else’s lifetime of data, I’ve realized that I’ve barely begun to think this through.

My sincere thanks go to Possum’s family for giving me permission to discuss this in a column.