In a New York Times op-ed last weekend, architect Michael Graves lamented the fading art of drawing—that which binds the dynamism of the human mind directly to an unencumbered stub of graphite. Like the malleable visions within our heads, a pencil in the grip of a practiced hand can flick at nuance and wrap a curve in a way that's neither inaccurate nor final—letting the line sashay just enough to let our minds decide where it best fits. CAD, on the hand, insists on rigid absolutes: sharp corners, perfect circles, and clean, defined lines that leave little room for impromptu invention.
Machines, Graves pointed out, are no longer handling only the prosaic tasks of design. CAD was once a tool to pump out accurate construction drawings and utility plans, but there now exists software that will automatically consider size, style, budget, and materials—as well as produce designs on its own. A drive through fresh sprawl leads you to believe that cyborg-designed dwellings will not only be built, but that they will, at some point, dominate construction.
Graves makes the case that automation is killing innovation within his trade. He very well could be right. But automation, when put to work correctly, can be a tool that engenders creativity rather than quashing it. Fretting about innovation lost to automation should be left to those like Mr. Graves: splendidly accomplished, unquestionably talented, and, in the world of design, quite famous. The rest of us should get on with the automation. When we do this, we give ourselves a chance to innovate anew.
All businesses, no matter how small, can benefit by introducing their workload to a few automating algorithms. When Aisle50 exited the accelerator Y Combinator a year ago, we were spending 15 hours a day in a California garage toiling over tasks that were the core of our operations. We parsed lists, sorted spreadsheets, and built deal emails by hand; our customer service was done by calling our database with SQL commands from a terminal window. We were unable to improve our product or innovate upon what we'd already done since once we flipped the on switch and the users showed up, we had to keep the lights on.
Luckily, we caught our breath. We soon directed energy and weeks of work toward vanquishing these necessary tasks through automation. A child can now look up anything and everything about each one of our customers on an intuitive admin panel. This allows each one of our 10 employees to easily rotate into and out of a customer service role—and they do. Our sales team can concentrate on sales rather than learning how to master join commands on multiple data tables. Our developers don't need to waste 15 minutes of their day sorting our customers by zip code to determine who gets which deal. All of these things now happen with no human effort. Automating these tasks may all seem rather obvious, but we're consistently surprised by the kinds of companies—big companies in some cases—that do things the hard way.
By wasting your and your employees' time on tasks that could be taken care of by an algorithm, you lose days that could be spent developing a new product or spent brainstorming how to fix a consumer disconnect with your marketing plan. When these rudimentary tasks get automated, employees are liberated to spend time pushing your product to be better and, ultimately, innovate upon your current paradigm. Innovation isn't just invention of the totally new, it's also about the stream of little improvements that make a product better than all other comers.
You don't need a corps of hackers to automate some of the dirty jobs that may now take place in-house. Menial tasks are easy to automate by simply shipping them out of house through a tool such as MobileWorks or oDesk, which provide marketplaces for outsourcing. We pay others to transfer names from scanned sign-up sheets we collect in our retail partners' stores to a spreadsheet we can easily load to our database.
More complex automation requires a larger investment, however. Facing these bigger costs, companies can fall into a trap of not automating by weighing benefits only in the short term. A manager looking at today's costs of automating a 15-minute daily task may perceive the price to be high: several full days of productivity from a programmer for a task that, after all, only takes 15 minutes a day.
But such a calculation is shortsighted. An easy 15-minute task each day rolls into more than a week and a half worth of work each year. At a salary of $75,000—call it $90,000 with benefits included—that time cost you $2,812. Twenty employees wasting 15 minutes per day on lame tasks costs you $56,250 every year. That's equivalent to another diligent, hard-working young employee or—if you're a small Valley-minded tech firm—a few years' worth of decadent free lunches.
The benefits of automation go far beyond cost. Now your best people needn't tithe away brainpower for repetitive jobs.
At Aisle50, we automated the process of logging deal redemptions from raw data fed to us by the point-of-sale software at our retailer partners. Instead of an engineer groaning through a daily 10-minute manual process to download and sift a CSV file, our app grabs the file from an FTP server, ingests it, and sprinkles the results into our own database. Picking off annoying tasks like these, one by one, eventually gave us enough time to closely examine, quantitatively, where we were losing customers in our checkout process and then consider how we could keep them in it. With time to spare, we got to tinker with small changes to our product that lead to large changes in our conversion rates. When work can center on interesting problems like these, sharp employees stay engaged and happy. Smart people, it figures, don't enjoy dumb work.
But smart work can be automated, too. Mr. Graves isn't wrong to worry about machines' effects on the purity of his profession. As I outline in my book Automate This, automation at the highest levels of some industries can lead us to nebulous intersections of art, efficiency and, on some level, humanity.
David Cope, for instance, has created sets of algorithms that write classical music modeled on the masters like Bach, Beethoven, and Mozart. As the music created by Cope's bots has grown more compelling, more haunting, more human, the University of California-Santa Cruz professor has become a polarizing figure. Some music scholars and composers have shunned him—one even walked up and hit him in the nose at a convention in Europe—but there is no denying that the music created by Cope's computer code can be moving and electric. As the science that Cope has kindled moves into other kinds of music, it's my belief—and the belief of those more deeply steeped in computer science than myself—that algorithms will eventually write a high percentage of the music played on Top 40 stations.
Music is a business, after all, and if a record company can employ an automated hit creator—or, as they're already doing—algorithms to sniff out the next big artist—why shouldn't they?
This goes back to Mr. Graves and his hunch that something won't be quite right if machines start designing, writing, and creating much of what we ingest and live amongst. But that day is coming.
The good news is that most of us are still trying to automate numbing office tasks and low-level development. Get that done and our daily work becomes more interesting—leaving us more time to ponder the tougher questions, much like Mr. Graves.
Christopher Steiner is an engineer and the author of Automate This: How Algorithms Came to Rule Our World (Portfolio | Penguin, on sale now). He's also the cofounder and co-CEO of Aisle50, a Y Combinator startup offering online grocery deals. Steiner was previously a technology staff writer at Forbes and the Chicago Tribune. He lives in Evanston, Illinois, with his wife, Sarah, and their children. Visit him at www.chrissteiner.com.
[Image: Flickr user North Charleston]