Mind The Gap
Once upon a time, my friend was developing SW for a new feature phone. There was a bug blocking the SW release for the production pilot run. Somehow no one seem to be able to reproducible beside a few on the factory floor. And all teams were pretty sure their code could not cause such problem. Late in the night, he could finally seat down to debug after talking about the bug for days. The office was quiet & his development board was kindly to repeat the error sometimes. This could be the night to nail the root cause down to unblock the release. As he quickly tested to rule out a few theories & re-hypothesized a few more, the morning was about to broken. Out of the blue, his VP stopped by. He looked very worry about another delay of the pilot run. It costs a lot to hold production lines. The VP asked: “A bug is just a few lines of code wrong, right? Let me help. Two of us can spot them quicker…” So they did spend the rest of the night looking for “the wrong code”. Unfortunately, they did not “spot” any that night. Maybe the bug was not in the code they reviewed, or sometimes 1+1 is less than 1 😭.
To be fair, the VP used to be very hand on. Only back then, the SW or firmware more precisely was much simpler, such as configuring & operating the HW. There was no a lot sophisticate functions implemented by SW. Today the world is a very different place with many smart things defined by SW. Mckinsey suggests it’ll be very tough for carmakers to keep up the accelerated growth of software complexity as users expect cars to get smarter with Over-The-Air updates. And the growing complexity could not be tamed by adding more people to the death march anymore.
To Collaborate Or Not To Collaborate
Mckinsey points out a great strategy as OEMs could join hands to develop an end-to-end SW platform. But such collaboration rarely happens even it’s obvious the best interest to collaborate, aka the prisoner’s dilemma. You’ll be surprised how often people can even be less cooperative under pressure. And in business, there is never a shortage of pressure.
So, how to make it collaboration worthy? Open Source is the sample first step. Because it sets no barrier for any player to adopt. It puts no player under no one’s mercy. Furthermore, it partly “solves” the knowledge sharing for collaboration. Because anyone can check “the source of truth” anytime. The key question is how to make the collaboration easier as the complexity grows.
Sharpening The Axe 🪓
Give me six hours to chop down a tree and I will spend the first four sharpening the axe. Abraham Lincoln
16 years ago, my old employer, TTPCom paid good money to equip us with a smart IDE, Source Insight to get the job done quicker. Even it’s in the company’s best interest for productive employees, I’m always grateful when companies invest so. After all, not all companies can afford that. Luckily today, a codebase can be even more accessible with online code search services, e.g. github.com & cs.android.com. However, it still takes a lot of brainpower to dig into different layers to see how things work & connects. It’s especially painful to root cause a complex issue on a codebase mixed with legacy & new code under pressure. Each time when I’m there, someone gives me a sharp axe 😉.
Seriously, I’ve been thinking about a powerful “chainsaw” for years. A tool to make it easier for anyone to see bigger or closer pictures of a complex SW system. Because the brainpower should be burned to gain knowledge & insight instead of processing data & information. With proper tools, the collaboration may be easier & suitable for a bigger scale. For years, I’ve been bouncing a crazed idea: Generative Super Vision SW Architecture Oracle. Next time, let’s scratch the surface.
Full Disclosure
The opinions stated here are my own, not those of my company. They are mostly extrapolations from public information. I don’t have insider knowledge of those companies, nor an EV expert.
