Kill Your Darlings

There is a lovely little phrase that most writers know, some writers appreciate and a precious few actually follow: “Kill your darlings.”

Short. Clear. Emphatic. Moderately violent. But somewhat ambiguous in terms of actual intent.

The intent of the phrase becomes somewhat clearer in the context in which it is used, which is the process of editing. Editing is critical to good writing (as I’ve come to appreciate somewhat later in life), but it’s a brutal process. Especially to those who invest themselves in their work, and see every mark of red ink on a page as a cut to their very soul. Never let it be said that writers aren’t also very often drama queens with delicate egos.

The phrase has been attributed to many, including William Faulkner, Mark Twain and Stephen King. King’s version of the quote is particularly gleeful, which is possibly unsurprising given the source: “[K]ill your darlings, kill your darlings, even when it breaks your egocentric little scribbler’s heart, kill your darlings.” That I can find, the earliest appearance can be attributed to Sir Arthur Quiller-Couch, as part of a series of Cambridge lectures in the early 1900s entitled “On the Art of Writing.” Although he indicated that darlings should very specifically be murdered, not just killed.

Editing is the judicious determination of what to leave in, and what to leave out. And what we should be excising, very clearly, are our darlings. And what, you ask, may those be? They are those choice, clever, delightful bits of wordplay which excite the writer to no end to have imagined, that do absolutely nothing to advance the story. They are the things that exist to further the author’s ego. They are—in the words of Quiller-Couch—”extraneous ornament.”

It’s extremely good advice, if difficult to follow. I love language, adore clever turns of phrase and appreciate well-crafted puns and plays on words. And I particularly delight in the moments that my brain offers up something amusing and appropriate. Even if its inclusion only amuses me. Because those are the darlings we are specifically discussing here. If words don’t serve the larger purpose of what you’re writing, they don’t deserve to stay.

But what, pray tell, does hundred-year-old writing advice have to do with project sponsorship? I’m so glad you asked.

Darlings, to put not to fine a point on it, do not only appear in word form. Sometimes they manifest in much larger contexts. Like projects. Which isn’t to say that we should be focussing on those aspects of the project which don’t serve a larger purpose—although we should. More particularly, sometimes the darlings that we need to be killing take on the size and shape of entire projects.

And this is where the phrase “kill your darlings” becomes extremely relevant. Because, quite frankly, by the time projects actual attain life, there is every expectation that we’ve fallen more than a little in love with them. And that’s where things get truly problematic.

As I noted in my last post, one of the roles of project sponsor is to sell the project. They are the champion, advocate and passionate defender of the projects they oversee. Given that projects consume resources and occupy space that could otherwise be occupied in the organization by other projects, this requirement for support is not insignificant.

It’s important to recognize that an essential component of selling projects is emphasizing the value that the project will deliver. That’s very often framed by highlighting the consequences if the project doesn’t go ahead. In other words, projects get sold by arguing vociferously for just how essential they are to the success and survival of the organization. We position projects as being critical and vital. We engage in rhetoric about how indispensable projects are. The more compelling we are in making the case, the greater the support, and the less likely it becomes that failure is in any way an option.

That’s a problem. It’s a very big problem. We know from research on project failures that the selling of projects involves systematically overstating benefits and underestimating costs. This strategic misrepresentation is not new, and it is not rare. Research into large scale infrastructure projects conducted by Bent Flyvbjerg finds consistent and persistent optimism in early forecasts that is specifically designed to make projects more attractive and more likely to be approved.

The hope that underlies this exercise is that once the project gets approved, the organization will be so committed to its delivery that the inevitable overruns and delays will be accepted and accommodated. Even more disturbing is the fact that this is a widely accepted strategy, and tends to work more often that it doesn’t. Project managers and sponsors both joke about the likelihood that estimates are going to grow, and the projects get approved anyway.

A project that I was involved in auditing a little while ago was a particularly good example of this situation in action. Even by the time the project was tendered, the timeline and costs had grown substantially from original estimates. The client, burned by previous project failures, had built in stringent penalty clauses in the event of delays. As a consequence, virtually every vendor the client hoped was going to bid chose not to do so. They went with the best of the few bids they did receive, which still was higher than the design forecast had suggested would be appropriate.

It didn’t take long for the troubles to start. The project manager was relatively inexperienced. There were numerous problems with the performance and reliability of sub-contractors. There was an enormous amount of political influence at play, with new requirements being identified, and key design decisions being overturned later in the process, requiring significant re-work. Material delivery problems led to delays. And the price tag kept going up.

In the face of mounting challenges, what did the organization—and particularly the project sponsor—choose to do? Keep pressing forward. The argument was that the project had come this far, and was so critical, that they needed to keep going. And so they doubled-down and hoped for the best.

When this happens, there are a couple of things going on that materially influence decision making, especially at an executive level. The first is what’s called “sunk cost bias.” It’s what happens when someone makes that case that “we’ve spent this much already; we can’t give up now.” Variations of this argument also include “We’re so close to the end; we just need to keep moving and get it done.”

The reason it’s referred to as a bias is that’s just what it is; we weigh the cost of what we’ve already spent, not what we will still have to do in order to finish. Astonishingly, even in the face of significant failure and challenge to date, there is an optimistic belief that problems will magically resolve themselves and the finish line can be attained. We are emotionally invested in what’s already happened, and we can’t rationally see our way out of that hole. Instead, we keep digging. And digging.

What feeds sunk cost bias is an aversion to loss. We hate failure, and are loathe to entertain it. And so we optimistically delude ourselves into believing that things can get better. Even in the face of massive failure and challenge, we invent ever-more-intricate scenarios of how hope, effort, a bit more money and a bunch of hard work can turn things around. And that leads us to a slightly different—but no less important—phenomenon that is formally referred to as “escalating commitment to a failing course of action.”

Yes, that’s a real expression. And it’s a real phenomenon. Escalation is the love child of loss aversion and strategic misrepresentation. And if that sounds complicated, messy and problematic, then you’re on the right page. We’ve sold the project on it being critical and vital to the success of the organization. We’ve since run into numerous issues and problems… But the project is essential… But overruns… But it’s crucial… But delays… But nothing. Escalation is a reinforcing cycle of ever-increasing hope mired in ever-expanding failure.

If that sounds hard to escape, you would not be wrong. One of the leading researchers into escalation is Helga Drummond. Her case study of the Taurus project is a notable—but not unique—example of escalation in action. A project to modernize the settlement of transactions for the London Stock Exchange, it was one of several projects being done, and probably the least sexy in terms of visiblity and impact. It was also a project where there was little agreement on requirements, and numerous political implications and consequences, both large and small.

The project started in 1986, and was estimated to require 3 years and cost £6 million. It’s urgency escalated massively after a 1987 stock market crash, but over the course of the next year expectations and requirements massively evolved. By the following year, the estimate was £60 million pounds, resulting in enormous pushback and resistance from member banks. A re-design was launched, which unhelpfully attempted to build consensus by welding together aspects of 17 competing solutions, at a cost of £50 million and a duration of 5 years.

Even while opposition was voiced quietly, and the market realities the project was supposed to address were changing, the project was seen as being so vital that no one dared publicly criticize it. The project continued to change, requirements continued to evolve, delays and overruns continued to mount. A consultants review assessed the project as still feasible, but with an anticipated delivery of 1993, a full seven years after the project initially started. By this time, costs were now estimated at around £90 million.

I could go on with a litany of problems and failures. It makes for a gripping—if depressing—story. Ultimately, though, the project was cancelled in early 1993. Incurred costs were £80 million, and outside securities firms had invested about another £400 million collectively in building their own software in preparation for Taurus. That entire investment was essentially abandoned.

How do projects get here? One step at a time. They are seen as essential, they are seen as inevitable, and ultimately they are seen representing so large an investment that to walk away would be unthinkable. The greater the sunk cost, the greater the resistance in abandoning the work. And yet, ultimately, that is often what needs to be done. The problem of de-escalating commitment, though, is you need to actually challenge the very case that got you here in the first place.

In other words, you’ve got to find a way to kill your darlings. What that requires, practically, is several things. First, you’ve got to be objective and realistic about project success, and the degree to which the project can actually be successful. This means tackling the very real problem of magical thinking. It requires challenging head on the notion that—despite overruns, delays and problems—things are better now and everything will optimistically turn out just the way we plan.

The second issue in de-escalating commitment is challenging the reason for being of the project. This is often a hard one for sponsors to grapple with, because it means backing down from many of the assertions that sold the project in the first place. They have to walk back the claims and promises. It shouldn’t be hard to build the case for cancelling the project, because the objective data should be available. What’s painful is making the argument, because it first and foremost requires arguing that the previous case was wrong.

Finally, there is the expending of political capital to make the argument heard, and to support the difficult but necessary decision of cancelling the project. In the case of Taurus, the president of the Exchange volunteered himself up as the scapegoat for the problems, offering up his resignation to the board as he made the case for cancellation.

Part and parcel of this is also making the argument of what to do next. This requires seeking alternative strategies, and being able to frame what is a reasonable next course of action. Sometimes, that might literally be to do nothing. By the time Taurus was cancelled, the operations of the industry had sufficiently changed that the compelling need for the project had essentially disappeared. That on its own is a difficult—but necessary—thing to admit.

Selling cancellation is made much more complicated without an alternate course of action being mapped out. It’s the essence of building an exit strategy, and it is best framed first by recognizing that if you were doing the project over, doing what you are doing isn’t the right answer. And so we need to conceive, from here, what the best course of action now actually looks like. That course of action can take lessons from the past. But it needs to ignore the pressure to keep going, or to take a similar approach, because of what has been invested to date.

Responding to issues and challenges is never easy. When the problem at hand isn’t part of a project, but in fact the project in its entirety, the challenge is that much greater. The hardest decision a sponsor has to make isn’t to proceed with a project; it’s to make the project stop. It’s always hard to kill your darlings. But every so often its absolutely essential.