Single point of failure as a design.
Today, I worked my normal day, some coding, some chickens, some walking the dog. Besides the internet seeming a little slow on the download side it was a pretty normal day. Until I got a message on Google talk chat plus hangout or whatever they are calling it today. And apparently, mobile phone service to a major provider is not working in my region. And I'd say it's a major outage with no clear answer as to when services will be restored. To me say one of 2 things, likely not under control yet or the news that supports the outage is not what you want to get out there at the moment while hundreds of thousands of customers who use phone text and data with close to 100% uptime, to the point where uptime and downtime is not in the vocabulary of a common customer.
But look, I can't make dinner plans or know exactly when my wife is in route, for others in potential more sensitive situations, this is terrible. So those with web application running in the cloud talk about disaster recovery plans. Not many are there yet, but multicloud approaches are doable and would have on demand cost which is nice and efficient. But for the rest of us we are locked in to the single point of failure. For years I held on to my land line, actual an inexpensive voip service. But let's be honest we will forget this moment that service is down and it also will soon pass.
I hope I am not jinxing anyone but what happens when there is no office to go to and there is no access to internet. Maybe as simple as taking off a few days until it passes. Let's call out single point of failure in a different phrase. Many teams are not able to pass the bus test. Which in retrospect is just a single point of failure. Stakeholders can huff and puff all they want but when I actually am not able to do something, somehow they find the ability to push the dates back.
And this leads me to my favorite conversation of late. The increase of the epistemic gap which is not going to be fixed by AI without the right SME’s, will create a single point of failure at many companies. Today it's a platform license that you aren't getting the value of, tomorrow it's a subscription to the tool where the knowledge exist. If a company had issues evaluating value and efficiency before than that will persist. If a company was good at creating single points of failure, and tech debt, this will likely expedite. It's still a people problem, for once the tech is perfect but it doesn't mean it fits.
Do your work now so this doesn't become you. You want to benefit from the rich tooling with skilled people who adapted to it seemlessly and new professionals that understand how to optimize historical set backs. Not a warning it's a promise - stop designing single points of failure now. They exist in and outside of technology.