Build no infrastructure, then we’ll talk

We’re in this strange ol’ time where it feels like so many aspects of our lives are maintained by ubiquitous, real-time connectivity to the internet — yet, the stability and trust in that connectivity, and the services that provide it, is slipping away at an alarming rate.

We’ve seen, over the past year or so, that as the economy has battened down the hatches, some much-loved platforms and services didn’t make it into the storm-shelter in time, and simply vanished.

And you don’t need me to tell you that there has been a significant change in ownership at one of the most popular social media sites on the planet, and with it, changes in policy and access to the underlying API’s that people built their livelihoods on. Do you think those people will ever trust a platform again?

Suddenly, some of the services that flowed into our homes as reliably as plumbing and electricity aren’t there anymore. Now what?

People have lost trust in the concept and reliability of the ‘platform’. That’s what.

So what to do, should we all convert a room in the house to be a datacenter so we can host our own tools and services? Should we buy a bunch of IPv4 addresses — if we can find them, and learn BGP so we can all become our own ISP’s? Should we start hosting our own email servers again? Maybe we should partner with our folks in our neighbor to create our own little mini-internets. No. Definitely not. Especially not the email thing.

We do need to have a serious discussion about what a ‘platform’ is to us, and how we interact with it — in terms of both how we connect to it, and what happens if it’s not around, either permanently or temporarily, for whatever reason. While I don’t think we need to go back to a bygone era of computing, I do think there are some lessons from the pre-cloud, pre-SaaS, ‘be self contained’, times that we can bring forward to get the best of both worlds.

That is to say, we build to take advantage of connectivity and cloud computing when its there, but we don’t build assuming it always will be. Let’s take the onus of maintaining high speed, reliable internet connectivity off of our users — because in some cases, they just can’t for reasons outside of their control.

It’s quite the shift in thinking, so we have to start somewhere. The first question to ask ourselves is — does this data really need to be shared in real time? A lot of things do of course, and thats fine. No point having real time telemetry of some industrial equipment critical for safety, and waiting a day to share it with a control center, for example. But a lot of things don’t, either. Payments, for example, can be reconciled asynchronously in a way that reduces the risk of fraud. Non-essential status messages can be passed to and from remotely located machines where no fixed connectivity exists.

The important thing — is that we build things in a way that the world doesn’t come to a grinding halt if the real time connectivity isn’t there for a hour, or a day, or a couple of weeks. We need the smarts of smart devices, we don’t need them to become dumb the second a cellular network operator stops operating when something breaks down, or US-EAST-1 has a hiccup.

Platform providers can help with this. I think the best platforms, going forward, will build in a way that allows people to be extremely modular about how they use them. Platforms can compete to be the first choice landing spot, rather than the one and only destination. That probably doesn’t sound all that appealing to the business folk who work for those platforms, but you’re selling to a bunch of technologists who have lived through the wave of vendor lock-in’s that didn’t end so well. They don’t wanna do that again, trust me.

And if you’re a software developer, build up the offline experience of your software to be as awesome as the online one. Don’t just throw an error — think about what you can offer your users — it’s not their fault something is down, well unless they work for AWS I guess.

I truly believe that ‘offline connectivity’ and resilience in the face of outages and disruptions are going to be the next most important area of focus in our connected world — and I can’t wait to see what people build around it.