Rant: A computer-science education suited for startup failure.
After many years of hard work inverting matrices and programming scheduling algorithms for that half-assed operating systems course assignment of yours, you graduate from university; snatching a 9–5 junior test engineer position from underpaid talent recruiters dumping you into a corporate cubicle with a bunch of other self-proclaimed industrial-grade engineers.
Housing, living expenses, and more paid for as a result of your long academic career. But the stable life was not good enough for you, and all you yearn for is to be freed into building novel product ideas into a reality with the tech prowess you’ve garnered over these past few years.
And so, you decided to make a startup.
You’re the only co-founder. CEO, CTO, CFO, COO; you had the right to claim all those roles, and thought you deserved a pat on the back. Server fees were paid with your remaining salary, the platform launches, and you went out to drink with your friends over your one-hit-wonder success.
The very next day, you wake up hung over only to realize that your spontaneous drunkard self got one of your childhood friends as CFO on board, and he has some bad news to tell you.
That one 150$/month server you bought on AWS for your NodeJS backend, real-time database, storage and React frontend couldn’t handle the influx of a mere 500 kids thrilled and entertained by your platform. The servers crashed, children started crying, and parents called in demanding why their kids favorite entertainment platform popped up a
HTTP Error 404.
You pop some pain-killers, pull out some distributed task queues libraries, whip out that good ‘ole algorithms-behemoth textbook back in your academia days, and get the platform back up with two additional servers leading you to your first tech. infrastructure behemoth ready to get some of those kids some of that Netflix entertainment.
Your parents ask how you were doing with your startup, and you shut them up with success stories of startups being seeded with the founders parents money hiding behind the 3000$ you “borrowed” off of their shared debit card for server fees and funny-ass-storage-as-a-service API fees. Oh, and you also move back in stating that you’re working on being the messiah of children television shows.
The platform goes back up.
Parents were happy, children were exhilarated, and your CFO started clocking in some news articles and commercials to be aired on Disney Channel.
Netflix got hold of the success of your platform, and the news of your Scooby Doo entertainment paradise gets a hold of a few hundred thousand families out there in the world.
But did I mention that your company wouldn’t break even until after this month was over? (the freemium model monthly fees thing was your CFO’s idea)
Servers crashed; again.
Those distributed queue libraries of yours in Python couldn’t process Barney videos fast enough.
Those livestreams to The Wiggles to over 5000 kids located in East Australia ate up all of your servers RAM.
Those $0.3USD machine learning API calls to M$crosoft Ass-zure got wasted due to sporadic message queue system failures.
To top it off, your prowess in failing to breathe and speak object-oriented programming got you into call-back hell with asynchronous error messages.
You children entertainment murderer you.
Your parents get funny calls from angry families about their kids not getting to see Timmy Turner from the Fairly Odd Parents, your CFO leaves you because he landed a job at Goldman Sachs, and you thought you dun fucked up.
There goes your hopes and dreams :c
You file for bankruptcy, working a minimal-wage salary looking back to how you still could have been the renowned children shows messiah.
It’s a pity you didn’t learn much from those internships apart from using open-source tools made out of a persons free time, huh?
Embracing Failure with Moores Law
Moore’s law is the observation that the number of transistors in a dense integrated circuit doubles approximately every two years.
It is also the observation that made programmers think that using 16GB of RAM and 100mb worth of disk space for
node_modules was a necessity.
That few kB used to launch a NASA shuttle into space was well worth those 32 different types of HTTP body content parsers bundled in your API libraries.
The point being though, is that the more internships, coding bootcamps, and universities embrace the use of these exciting rapid prototyping technologies which reduce potential for a programmers errors, the less prepared programmers are in amidst the time that comes when a startup needs to technologically scale.
Rapid prototyping frameworks were made to either test ideas quick, or made to cater for businesses that have always had serious cash from the very beginning. The real world costs money, and these frameworks were not made to make the most out of the server resources an unfunded startup has to spare.
The next time you day-dream and figure out that next great big startup idea of yours, get your hands off those scripting languages and start doing some work in getting your MVP’s scalable from the very start.
Lines of bleeding-fast C++ or Go or Rust code, or even Erlang code made simple with tons of well-maintained open-source servers/libraries out there, with a custom ZeroMQ/nanomsg/etc. inter-server broker is gonna get you ready for the time you start accepting +1mil HTTP requests/network packets a second with minimal server costs necessary.
Get your ass off of so-called developer productivity trends and work with tech. that’ll enable you to be the next big business maker. Even a 6 year old could understand distributed systems done in C++.