Even nerds don’t want to run their own servers
This is a quote from Moxie Marlinspike’s post My first impressions of web3. This article is great critique of web3. He makes a number of good points, but what I found most interesting was his assertion that people do not want to run their own servers. And I cannot stop thinking about this and not b/c I think he’s wrong, but b/c I think these challenges can be overcome.
Moxie doesn’t go into any detail but here are some of the problems.
Networking
DNS
Most servers, and the ones Moxie is talking about, are ones that are meant to be interacted with outside of your own home. Most of us in the U.S. do not have a static IP, so you’ve got to figure out how to get the domain name resolved as the underlying IP address changes. There are number of ways to solve this or work around it:
Use the dynamic IP address and update your bookmarks as it changes. Ok, no, this is not a real solution.
Use dynamic DNS service. This is a service that requires you to run a small piece of software that updates the IP address that your DNS entry points to. The best place to do this is on your router, but many routers do not support it or charge for it. Of course, you’ll need to setup your own domain name for this too and setup the service such that it’s able to change the DNS entry for you. This is all do-able if you’re reading this, but no one wants to go through the hassle.
If you only need access from your personal devices, you can use Tailscale to create a virtual network between your devices and your server. You can use the handy “magic DNS” feature such that tailscale will resolve a hostname to an IP address of your choosing. I.e., on my devices “v.net” resolves to an address on my local (tailscale) network. The downside is that these services cannot be accessed by people who you have not authorized, so will not work for everything that you might like to host.
Security
If you’re running services over the public internet you will likely need to secure the connection via https. While it’s possible to create free certificates that will be trusted by all browsers without warnings for free (thanks LetsEncrypt!) this is another area of pain. The certs will need to be renewed occasionally. This is a common enough failure point in the enterprise. You can figure it out but no one really wants to.
If the service is only for your own use, Tailscale solves this problem as well as you don’t need https at all.
Storage
Storing data is easy. Making sure you can survive a disk failure is something else entirely. You can use some sort of RAID, but if your house burns down, you best have an off-site copy too.
There are plenty of backup solutions of course, but setting it up, ensuring that it continues to work, and restoring from the backup when it does fail is all work that no one really wants to do, even those of us that do this kind of thing for a living.
There are companies that are attempting to solve this problem. One promising solution is Functionland which ran an indiegogo campaign recently for their hardware solution and software platform. The basic idea is that everyone with one of these boxes stores everyone else’s data so that it can be retrieved again if your box fails. This solution is squarely in the web3 space, backed by a token which aims to incentivize app developers to create apps that will run on the device.
Disappearing tech
So, yeah, self-hosting is too much work. I’d prefer not to have all my data backed up to Google but it’s just the easiest option. Many of these problems are tractable and we will see companies come along and change the landscape in the future. Tailscale is an amazing example of this. Perhaps Functionland will chip-away at the storage related problems. I’ll be watching.
I do hope we will all be self-hosting in the future, but if so the experience will have to more like installing a Nest thermostat or setting up a router or a new TV. Sure, none of these are experiences are particularly pleasant, but you follow the instructions, get it setup and forget about it.