Software Hasn’t Come Close to Eating the World
The programs that were supposed to be fun, smart and make our lives easier are so disappointing.
Bloomberg, October 18, 2017
Software is so disappointing.
It was supposed to — in the words of one sage — eat the world, the implication being that it would be modern, brilliant, liberating, hip and fun (assuming it didn’t make you and your job obsolete). Instead, it’s pretty terrible.
It isn’t that we are hurtling too fast into the future, but the exact opposite: Why is the future taking so long to get here?
Some of the blame goes to the programmers who write the software, along with the marketers who endlessly hype and oversell the tech future. Too much time and brainpower is devoted to designing improved sexbots and not enough on making a working robotic personal assistant. There are billions of dollars to be made solving these problems for the first company that figures out the latter.
Forget Moore’s Law; as an alternative, consider Ritholtz’s Law of AI: our impatience grows exponentially with each step closer to functional machine learning and artificial intelligence. It leads us to realize that what we want is even further away than once imagined.
Contrast that with the words of Marc Andreesen, who wrote in a 2011 op-ed under the headline “Why Software is Eating the World“:
Six decades into the computer revolution, four decades since the invention of the microprocessor, and two decades into the rise of the modern Internet, all of the technology required to transform industries through software finally works.
“Finally works”? Six years after that was published I can say with confidence: Maybe it works, but it sure doesn’t work anywhere near as well as it should. My list of software fails goes back a decade, and almost none of these issues have been resolved. To cite just a few examples.
Apple’s Siri: A business trip last year brought me to Italy, and I spent a few extra days in Rome. For almost 3,000 years, the city has had a global influence on art, music, architecture and, of course, food, especially the pizza. Back home, after hearing that an authentic Roman pizza shop opened a few blocks from my office, I immediately headed there. It didn’t disappoint — the pizza was delicious. I took a few photos, and used Siri to tweet them with the caption “Legitimate Roman pizza comes to New York City (37 & Madison).”
Only that’s not what dumbbot Siri heard:
Legitimate woman pizza comes to New York City (37 & Madison) pic.twitter.com/lm9jTURVF9
— Barry Ritholtz (@ritholtz) October 9, 2017
It is my fault for not looking more closely at the tweet before hitting send. But more importantly, why should I have to?
It goes without saying that there is no such thing as “woman pizza.” On the other hand, Roman pizza is a real thing. The way that Siri rendered my spoken words was off by one letter, yet it wasn’t smart enough to correct it. More to the point, “Legitimate woman pizza” is nonsensical. What’s more, I was located in front of a store that literally has “ROMAN” and “PIZZA” in its name on the sign out front.
It is not simply that Siri is useless, but rather that all of the tools for making it useful are right there. Too bad Apple Inc. doesn’t have any spare cash lying around to perform the needed research to improve this.
Alphabet’s Nest: I like Alphabet Inc.’s (Google) automated thermostat system for my home air conditioning and heating. I only wish it was smarter.
The schedule you set for Nest is kept on Alphabet’s servers. You can access it from anywhere you have internet access. Using the app or a web page, you can check on monthly electricity savings, set it to “away” when you’re on vacation — all very neat stuff. That is, until the power goes out, even for a second. Then Nest resets to default mode, and you have to reprogram it from scratch. This really should auto re-populate from Alphabet’s servers to your custom settings.
Uber’s arrival estimates: I recently decided to use Uber Technologies Inc.’s ride-hailing service instead of driving or taking a train to head into New York for a dinner that I knew would run late. The app gave me an expected arrival time of 45 minutes. Given that, my wife and I request the car 45 minutes before we wanted to depart. After I placed the order, the Uber app indicated that our car would arrive in nine minutes — before we would be prepared to leave. This could subject us to ridiculously high waiting fees if we weren’t ready when the car showed up. I called the driver and told him he was running way ahead of our schedule. He advised me to cancel, which I did. A bit later we request a second car; it arrived and we left. Then Uber hit me with $5 cancellation fee, based on its protocol that you have two minutes after placing an order to cancel without cost.
Wait a second. Uber’s software should have figured out that it gave me bad data initially and that the driver and I texted and spoke on the phone and that I followed his advice. This wasn’t my fault. It was poor asset management on the part of Uber. An intelligent software agent would never have billed me for the cancellation.
The net result: repeated emails back and forth with customer service, and a frustrating experience that eventually led me to: a) like the company much less; b) get a $5 refund; and c) open a Lyft account.
I have no doubt that the future is coming, and that eventually, Andreessen will be proven right. I just wish it would hurry up and get here already.
Originally: Software Hasn’t Come Close to Eating the World
