The specter of failure and the specter of success

The disastrous launch

It’s been a warm winter, but the palm trees are nonetheless shrink-wrapped up and down Atlantic Avenue. Virginia is not their native environment, and their winter discomfort is the price we have to pay to maintain the fiction that Virginia Beach is a summer paradise.

It’s day four of Tiber, Thursday afternoon, and the president has flown back from Atlantic City to deal with an office in absolute disarray.

Things are not going well at our logistics company. We’d done a hard cut-over on Monday morning, and every freight bill is now in Tiber for processing. Of the 600 bills that come into the office every day, we were able to process only 13 on Monday. Tuesday wasn’t much better: we batched 55 more.

We had not done any testing at scale, and now the server has been crashing several times a day. We had not done any training, so the audit department is in a state of undisguised chaos. This is all uncharted territory for me: I am 26 and Tiber is the first thing I’d ever built that actually mattered to anyone.

I’m at my desk, up to my elbows in bugs and fully absorbed in the Java files in my text editor, when one of my co-workers barges in. The president has assembled all of management in the conference room and has announced that we are abandoning Tiber.

I tab over to my terminal, commit my diffs, and run to the conference room, saying, “I'm here to stop you from doing this.”

The president replies, “It’s Thursday afternoon, and we’ve only batched 600 bills all week. My company is going to go out of business, and I can’t let that happen. We have to stop.”

I pick a number out of the air. “Fifteen hundred bills. If we batch 1,500 bills by the end of the week, will you agree not to go back to the old system?”

This gives him pause, and looks at his watch, as if to confirm that it is, in fact, still Thursday afternoon. “Okay, but we're at 600 ...”

I muster up the confidence to finish his sentence: “... and we’re going to batch 900 more by close of business tomorrow. Just” — I try to think of the most reasonable-sounding way to put this — “put me in charge of everything.”

The president turns to survey the terrified horseshoe of middle managers in the conference room and asks who thinks this is a good idea. Hands go up. The president says, “The office is yours.”

I think, maybe this is what The End of One’s Professional Career looks like.

The original idea

Tiber wasn’t even something that anybody else really wanted, but I knew that we needed it, I could build it, and it would be great.

Our clients manufactured large objects such as compressor pumps and yacht windshields. We didn't make anything or move anything ourselves. We were just an office of people making things happen with computers and phones.

The way we’d been doing our entire workflow was backwards: I could see this. Every day, 600 freight bills came in. We audited them, corrected them, marked them up — all on paper and with sticky notes. Then we keyed them into an MS Access 97 database and sent them out.

What we should do is data entry first. No, wait, the first thing we should do is scan the paper bills, and then the computer will be in charge of everything else! We’ll have complete visibility into every bill in our office! Most of our carriers have electronic invoicing! We won’t need to touch the bills at all.

I was in the president’s office making my case, waving my hands excitedly. He was regarding me with a bemused expression, but I knew he would come around. Our process hadn’t changed since the company was two guys working out of a living room, and when I got the job I had to request that the Norfolk Public Library take their MS Access 97 manual out of cold storage. It was time for a change.

The warning signs

We knew that transitioning to Tiber would be no picnic. The audit department was understaffed, and they’d been falling further behind for months. We didn’t have a great way of quantifying the size of their backlog, but the height of the piles of unaudited bills was increasing, and every so often an auditor, to her horror, would find a forgotten stack of pending bills in a desk drawer somewhere. Needless to say, no resources had been pulled off for testing or training.

In the week before the scheduled Tiber cut-over things looked pretty grim. The launch date had already been pushed back a week, but nothing more had been accomplished. My boss, the general manager, asked us, “What would it take to actually implement this project correctly?”

After an uncomfortable silence he answered his own question: “I'm thinking we would need to hire four or five people just for testing, and we'd need to push back the roll-out by two, three months?”

More uncomfortable silence, so he finished his thought: “We all know that's not going to happen. So let's start on Monday.”

“This is going to be the worst week of our lives,” I said.

The new economy

In the old workflow, auditors would take one invoice at a time and verify the zips and weights against the details on the bill of lading. They’d verify the calculated price against the pricing agreement and the appropriate carrier’s pricing application. They’d call to correct any discrepancies. Finally, they’d mark the bill up and add the coding.

With Tiber managing everything, an auditor would work through all the bills sitting in her stage of the assembly line. She’d have a carrier’s pricing application open on one monitor, and then Tiber would feed her each of that carrier’s bills for price run. Fifteen seconds per price run. As soon as she clicked the submit button she’d see a new bill, and Tiber would keep everything flowing downstream.

I’d written the code to perform price run automatically for all of our major carriers, so Tiber would only feed the bills from our lower-volume carriers to the auditors. It was impossible for an auditor to get a sense of where the bills were routing; all she would see was an endless queue of bills on whatever screen she was in charge of.

It was already an industry full of tedious jobs, and I’d made them vastly more tedious.

Once upon a time, computers were a tool that we used to get our work done. I’m part of a generation of bright-eyed technologists that have flipped that relationship on its head. Increasingly, the humans are a tool that the computers use to get their work done. The humans, in Tiber, were only ever a fallback mechanism to do the piecemeal work that Tiber itself hadn’t yet automated.

The fantastical interlude

When I try to explain my profession to my kids, I tell them that Harry Potter has it completely backwards, with its colorful robes and dark wizards. Real magic works nothing like that.

In real magic you descend into the twin kingdoms, Terminal and Text Editor. In these colorless landscapes, form and substance fade away, and your only tools are symbols. They’re the same symbols that you've seen your whole life — letters and numbers, dashes and braces and suchlike — the symbols that everyone else uses to jot grocery lists and text their mom, and otherwise do whatever people do.

But in these monochrome kingdoms, symbols have an unexpected energy to those who study them. If you've studied the appropriate arcana, you can arrange them to unlock a force with such power, a power that can benefit our restless world, but also a power that is capable of tearing the fabric of our society, as any cursory glance at a newspaper — if you can even still find one of those — will reveal.

In real magic the dark wizards are sometimes also the friendly wizards, operating at the vertiginous height of their best intentions.

Six months later

It’s summertime, and the heat is rising in waves off the sunbathers at the adjacent beachside retirement community who lounge, shirtless, not 15 yards from my desk.

The semi-retired founder of one of our largest clients is doing some sort of road race from Mongolia to Berlin. He's driving a '47 Buick convertible, and he lost his rear bumper outside Moscow. It has become our responsibility to deliver to him a replacement bumper, and this task becomes the major preoccupation of our upper management for the better part of a week.

Our international department smirk that they'd be able to get the bumper to the airport in Russia, but they do not have any contacts in the Muscovite courier network. So our customer service manager volunteers, with unnerving enthusiasm, to fly out to Russia himself and deliver the parcel by hand. He's already researched weekend hotel rates. He casually mentions that if he flies first class he may be able to take the bumper as a carry-on item. The customer service manager is an ex-Marine, and the bumper, when wrapped, looks like a Stinger missile.

The entire bumper contretemps I take to be a good sign. It's a signal, at least, that the rest of our various office crises have reduced to a simmer. Tiber has been in place for six months, and the kinks are mostly worked out. Our president has started to grow expansive. He's talking about the future, thinking long-term. I have just negotiated a huge raise and the guarantee that I be allowed to telecommute from Europe for six weeks every summer.

The specter of success

This is typically the part of the story where people start losing their jobs because of technology. It didn’t happen that way exactly, even though there was less work to do, and what was left could be done more quickly by people with less training and experience, working anywhere in the world.

I had hoped that our middle-seniority auditors would be freed up to join our nascent customer service department. Instead, they ended up all quitting in the space of three weeks. Turnover had always been high, and there were other factors at play, but the introduction of Tiber certainly did nothing to help.

We ended up promoting our entire data entry team to audit overnight, and we hired a new data entry team in India, which I figured out how to do after Googling, “India outsource data entry.”

In the end, nobody was freed up for higher-order work. Nobody had a lightened load now that the system had mostly taken over. Lots of people got moved up one rung of the corporate ladder without seeing much impact on their paycheck. And our customer service department was still just one person, now en route to Moscow to deliver a rear bumper.

I backed into becoming the data entry manager after hiring that team in India, and then I was the audit manager, and then the VP of something. Productivity went up 78% but morale was a variable I didn’t have an answer for.

By this point I was spending my summers in Sweden, where the government finances 500 days of parental leave per child. The time that I was, increasingly, not spending in the office was time that I was spending with my burgeoning family. I certainly have no regrets about that, only guilt about the circumstances that afforded me that luxury. Things kept getting better and better for me, but not for the co-workers whose jobs I’d completely upended.

The cliffhanger resolution

I’d had a decent hunch that we'd be able to hit 1,500 that first week. There was cause for optimism in the math: after two essentially lost days, we’d put through 300 bills on Wednesday, and we were on track for 450 for Thursday. Seven hundred more on Friday seemed achievable, but only if we focused on the bills we knew we could finish.

I already had some ideas in mind when the president gave me the carte blanche, so I started talking fast. Data entry needed to stop entering any bills that came in without all the backup. The receptionist could do price run between calls. The auditors should stop bothering with bills that came in wrong, were missing any information, or looked difficult in any way.

When we hit 1,500, Friday evening at 7:00 pm, we were all gathered around the desk of the auditor who polished off the last bill. As it slid into the batch, everyone let out a cheer. It was late, we were groggy from stress, and the fractures in the office were as large as ever, but we were radiant with pride.

The next week we batched 3,000 bills, and 15 years later Tiber is still the backbone of that company.