Whoa! What a ride! It all started with a "meh, I hate organize cards" and look where we are!
A bag full of experience
This year has been very useful to truly understand the whole "card sorting" niche. I was able to talk with customers first hand and truly understand their needs and requirements.
I learnt way too much about the physics of cards sticking together and how to overcome that.
I developed an eye on how to spot design issues and double check materials quality before using them in production.
How to build a fail-proof recognition engine that runs in the cloud.
Develop, plan and deploy software and hardware updates for customers scattered around the world, from New York to New Zealand.
And of course, you know, how to deal with the global pandemic that put all of us in danger and (from time to time) into lockdowns.
The law of large numbers
One lesson learnt was that large numbers suck. They are subtle and it's easy to ignore it, but they will come back and bite your back.
Let's make a very simple example.
Let's say there's a 50% that I'm going to give you 20$. You would be quite happy because half of the time we play this game you would get some money.
Let's raise this probability to 99% or 99.9%. Now you're very happy because you feel you're almost going to get some money every time we play this mind game.
So 99.9% feels like a trusty value, right?
Ok, now let's apply the same value to some kind of operation, say picking up a card and deploy it to a safe spot (sounds familiar, isn't it?).
So 99.9% of time everything will go fine, the remaining 0.1% will cause some kind of troubles.
It still feels nice, isn't it? At the end of the day, in the real world, every system will have a failure sooner or later.
But let's crunch the numbers.
If we apply the probability of 99.9%, it means that 1 in 1000 cards will cause issues.
In a single sorting batch, you can sort from 1000 to 3000 cards (more on that later), so we can expect that we're going to have from 1 to 3 issues on each run.
All of sudden, we understand that accepting a failure rate of 0.1% is not an option.
So what?
The need for a new version
People following me on Instagram or regularly checking the site should already got it. There's a new version coming up pretty soon (few weeks if everything works as planned).
So, what's the difference with the previous version? Well, nothing changed but everything is different.
Main features are the same (actually the same software is powering the current and the new version), but there's a focus on increasing the reliability of the whole system.
The amount of 3D printed parts is fewer than before, all core components are made in steel, a nice cover is protecting the electronics and finally all the cables are well organized.
As always, Magic Sorter is meant to be fully serviceable by customers: no glue, no soldering, no permanent or custom made components.
You'll only need a screwdriver, a pair of pliers and a set of Allen keys to completely disassemble and repair the machine. Everything is made with off-the-shelf components, so you can easily find replacements (except for the steel parts, but if you're able to bend them... well, I guess you're facing a bigger issue).
Stay in touch for further news!