local-first

A little late to the scene, I came across the Local-First manifesto while researching CRDTs (conflict-free replicated datatypes) for work. There's a couple of different sides to local-first that I find appealing, and they mostly stem from my vision of the future of computing from when I was a kid. I was lucky enough, before I was 12, to be able to plop myself down in front of a TV, surrounded by pillows, snacks, and sugary beverages, laptop on my lap, and hack away at all.bas, the source code for all my "programs". If I remember correctly, it was quite a while before I understood variables, loops, or the other fine things in software. Somehow, that didn't stop me from building things. I was sure then, and continue to believe, in a future with ubiquitous computing. I was on the verge of something awesome. A few expectations were key to that future.

This is not yet a given for all people, and there's even a widening trend in modern science-fiction to incorporate resource scarcity as a plot mechanism, but Pew Research puts US smartphone ownership at 85% in 2021, up from 35% when measured 10 years earlier.

I imagined this evolving to portable form-factors that, in most cases, didn't require hands. Certainly small, but probably wearable. Clearly, the early models would look like Pip-Boys, but eventually we'd have something closer to KITT in a Communicator Badge.

What we've got today is a strong start on the underlying technology, but the form-fact is still clinging to built-in displays. This will keep the devices (and their power requirements) larger than I imagined. Watch form-factors are getting interesting -- a pairing of watch and headphones, with the option for wireless display connectivity, would be very exciting. In fact, without the wrist-band, an Apple Watch is practically there. Apple and Humane seem to be on the right track.

I've been pursuing this, actively, since about 2010 when I concluded that a ThinkPad, running linux, probably had the horsepower I needed for software development. I have long been a console gamer, so I don't get caught up in the upgrade cycles or system requirements in that space. If I choose my development platform wisely (and lean on build-tools that aren't constantly rebuilding the world), a laptop today is more than capable of running all but the most compute-intensive workloads, like ML training or video rendering.

For years, I would select laptops to fit easily with my home cabling setup, specifically to allow me to switch between work and personal devices quickly and seamlessly. For a good chunk of time, this was ThinkPads on docking stations, followed by a gradual switch of everything to USB-C/Thunderbolt. Dual monitors, wireless keyboard and mouse, and the rest of the desk are just along for the ride. In my latest incarnation, I've mounted small cable trays and a laptop shelf under the desk, fitting the nearly-identical MacBooks I switch between.

In a few ways, the smartphone is starting to fit the description of "special-case peripheral". My laptop very easily tethers to my iPhone to provide internet when out and about, and I can use it as a camera or microphone should the need arise. It can also act as a wallet, and even an arbitrary RFID source.

Something that has yet to take off that I find particularly exciting is the idea of a handheld (or smaller) primary device, which extends into different form factors as necessary. The First has neat concept of on-person primary devices with an AR interface (glasses), and Caprica had tablets that appeared to be foldable sheets of paper, that also could be set on a workstation and sync seamlessly.

Ubuntu Touch has been chasing this for several years now, but appears to be stalled finding an ideal hardware partner. The software is exciting and robust, but has a limited audience based on community-supported hardware platforms and the expectation of a tinkerer as the primary installer. A surprising contender in this space is the Nintendo Switch -- a reasonably sized portable display, with a docking station that scales up to whatever monstrous slab of pixels you have in your living room. Nintendo doesn't get enough credit for it's work in this space.

Requiring a concept of "online" to access information (especially real-time info or communications) is one of the principal concessions I allowed for as a child; summers full of walkie-talkies with their roughly across-the-street range made an imprint. My expectation was that range and coverage would continue to expand with society, until a point that the only folks who ever saw gaps in connectivity were suffering legitimate device failure, or being an explorer.

Being mostly online, across long distances, likely means that centralized computing is going to be problematic. Offline data and compute will be necessary, which will require a shift away from the cloud-based applications we've converged on in the last 20 years. Fortunately, these tactics also work over short distances, and can be vastly more efficient (in terms of infrastructure usage and associated costs). This is where local-first concepts really come into play.

In part 2, I'll discuss how many local-first concepts can supplant existing products, while also touching on some of the big questions remaining in how we currently think about connected applications (like identity).