Not-So-Infallible Automation
Many modern comforts and conveniences are driven by automation. I manage my bank accounts and pay my bills through a series of scripts and databases, with rarely the need to interact with another human or walk to the mailbox. Thanks to self check-out stations, I can get in and out of the grocery store with ten bags of Keeblers without exposing myself to a cashier who smiles and says, "You really like cookies, huh?" leaving me to mumble that they're half-price and I'm stocking up. And since the invention of massive online stores, we no longer have to drive for hours looking for a tri-tip screwdriver that does not seem to exist in any hardware store in the state of Indiana.
But the world of technology is not all sunshine and roses. The great drawback is that, when companies expect computers to replace the work of humans, they often hire cheap staff to support it that doesn't have a clue what they're doing and can't think beyond hitting a set series of buttons. There was a time, when I was but a wee college student, that if the wrong price for snow peas came up on the register, the cashier would send someone to check the sticker on the shelf and change it manually with apologies. But over the past couple of years, every minor pricing dispute I've had ends quickly with a deer-in-the-headlights look and a, "But the computer says they're $3.99 per pound." Then they continue to scan and bag the item without another word, because nobody could possibly argue with the database.
Sweetie and I have been tearing our hair out over a similar situation with our apartment complex. A new company bought the property in December and, in the classic fresh-management tendency to make their mark by overhauling everything for "progress," forced the office to replace their old workflows with a computerized package. The simplest of tasks suddenly became impossible. I stood in the office for fifteen minutes once waiting for them to find and print a particular form. The staff kept a package for us in the back room for three days without a word, and when we asked why they hadn't told us it was there, the manager blustered that she was powerless and it was the UPS man's job to do that sort of thing. Breaking the pre-programmed routine and picking up the phone for five seconds was simply inconceivable.
These were minor inconveniences, but the real trouble started a few weeks ago when we turned in our papers for the 2012-13 lease. The girl at the desk accepted them with a smile and tossed them in a folder, without telling the almighty computer that we still lived here. I paid our rent for August and went about my business. Then, at the end of July, I found a notice on our door saying that we must sign our lease, or they might rent the unit out to someone else. When I called, the woman who answered (who I believe was the manager again) begrudgingly admitted that they had the papers and would fix it. They fixed it alright—and charged us rent a second time. We came back from a weekend away to a notice on our door saying that we had $510 overdue, plus $88 in late fees. I called to say I had the payment receipt, bank statements, the works, and I would "like" very much for them to remove the erroneous charges from my account. A voice mail the next day confirmed that they had, indeed, charged me twice by accident.
"Don't worry about the ledger that you got. Everything is fine and we will get this taken care of."
Naturally, a week later we had another notice on our door telling us that our late fees had now accrued to $136. According to the threatening letter attached to the first bill, a "written three-day notice to leave the premises" might soon be on the way, and if we fail to pack up and leave our "paperwork will be turned over to the attorney and filed in court." There is a real, absurd possibility that we will be served eviction papers by a machine.
Sweetie and I aren't worried, because even if the people in the office can't figure out which buttons to click, we don't have android judges (yet). And any flesh-and-blood one will take one look at the documentation, maybe listen to our safe-guarded recording of "Kristen" admitting their mistakes, and verbally box their ears.
A lot of modern publishing, especially for small presses and self-publishers, relies heavily on automation. Admittedly, it provides great opportunities for people once locked into a slow-moving, carefully controlled bureaucracy. But the freedom afforded by technology also comes with the usual pitfalls. In my short and not-so-lucrative career, I've already run into problems with every distributor I've used. First there was that little incident with CreateSpace, in which they warped my cover for Bubbles Pop and then insisted it was my fault. Just this month, one of my short stories was stuck in processing on the Kobo for an extraordinarily long time. My first email after day three went ignored, and my second after an entire week received a terse block of copypasta in response: "We are aware of an issue that a small percentage of our Writing Life authors have experienced with certain books not being loaded onto the site within the expected timeframe." They eventually fixed it and followed up with a second, somewhat nicer block of copypasta explaining that they weren't fully prepared for the rush of publication requests they received from launching Writing Life.
Two days ago, Barnes 'n Noble lost its distinction of being the only distributor that hadn't raised my blood levels of stress hormones. I republished "The Arrangement" and "The Mistress" under a new pen name now reserved for any stories with explicit content. Though the metadata was changed easily, the system did not seem to accept my new cover images. The records went live proudly displaying one name on the web page and a second on the cover. In my email to customer service, I was careful to say:
The jpegs fit the sizing guidelines (all sides between 750 and 2000 pixels, approx. 200kb files). I've tried re-uploading them several times in different browsers under different file names, but they just won't go through.
The response?
Please be aware that our system only accepts a .JPG file (or a .JPEG file) with a file size between 5 kilobytes and 2 megabytes for cover image submissions. Also, the sides must be between 750 pixels and 2000 pixels in length. Make sure that your cover art image has height and width (in pixels) within these bounds. If it does not, then you must use image-editing software to "stretch" or "shrink" the image, maintaining the ratio between height and width.
Gosh, thanks, "The PubIt! Team." It's so lovely to know there's still a company out there that reads emails, instead of shooting off useless form replies that don't address the problems at all.
Fortunately, it was only a propagation delay, and the new covers appeared by themselves after a day. Only a few hours later, when the same stories went live on Amazon, I discovered that I couldn't remove them from the "T.K. Marnell" author page, because it was against company policy to allow writers to choose which titles were attributed to their names. I couldn't even delete the "About the Author" section, because the form just spit back instructions to "Please provide content for review." I had to make a special request to remove them and wait for them to decide whether my intentions were "appropriate."
All of these anecdotes, at least, relate to tasks that should be relatively simple to automate. Paying bills, distributing digital files...you shouldn't normally need skilled labor to process these things. It's only a problem if you don't have competent people as backup when something goes wrong. Where it gets really ridiculous is when people expect software to be an adequate substitute for jobs requiring higher-order reasoning. There are downloadable programs out there that are supposedly able to proofread and develop your manuscripts just like a human editor. How do they accomplish this? By counting the number of times you use certain words and phrases, and by checking your passages against a dictionary of clichés. I've seen an outpouring of gratitude from some writers who've used these programs to "fix" their writing by isolating and deleting the offending idioms and adverbs. Yeah, that's all there is to it. Who needs readers these days, really?
Natural language processing is very difficult, if not impossible to do well. Especially for a cobbled-together language like English. Case in point: Amazon's new review snippets on product pages with "X reviewers made a similar statement." Exactly how do they judge which sentiments are similar? And on top of that, sort out which are most popular? Anyone who's dabbled in NLP will tell you that in our messy, redundant, metaphor-riddled language, structure != semantics. I've heard of writers whose books tanked in a single day thanks to the algorithm picking out the negative reviews to highlight, whether they were representative of the majority or not. Or you get fun results like this:
The most popular insight into Harry Potter and the Sorcerer's Stone by J.K. Rowling? "I read the book Harry Potter and the Sorcerer's Stone by J.K. Rowling." (1,244 reviewers made a similar statement).
Don't get me wrong—computers are great. I spend most of my life attached to one. But they're just tools. They're information storage and processing devices. No more, no less. You can't expect them to take care of themselves and us. Or you can, if you're cool with the consequences....
The machine knows!
No comments