September 5, 2019
Changebloga few thoughts on a friend’s podcast appearance
My friend Adam Jacob was guest on the Changelog podcast last month, and I totally missed it. Fortunately, a client fresh off a successful exit, who apparently just can’t take a break, happened to listen in, and sent me a link.
Mentioning me at all was not expected. But Adam said some of what I’d expect from there, like “I think Kyle is wrong.” Plus some absurdly flattering things, utterly unfit to reproduce here.
Adam and I will have lunch sooner than later. But I’d like to go ahead and push back on two themes I heard cutting across his ninety-minute appearance.
I think Adam very rightly criticizes tech businesspeople who focus too much on conversion rate and deal-loss aversion. But I think the same criticism applies to Adam’s own focus on addressable market. Rather than quit the cult of conversion rate to join the cult of addressable market, we should admit that the funnel model is mostly a descriptivist anxiety pill for managers staring down business uncertainty, and not a framework for deriving tactics that apply across all of software.
It’s just too simplistic. Your “addressable market” is everyone who might buy what you’re selling. You “convert” those maybe-woulds to paying customers at some rate of success. Paying customers return the money you spent producing your offering and landing their business, plus a profit, your “customer value”. Thus your business fortunes track:
Addressable Market × Conversion Rate × Customer Value
Adam rightly emphasizes that the goal is maximizing the product, not any particular factor. Higher factors contribute to a higher product, of course, but overemphasizing any one can get you stuck at a local maximum when you trade off the others. You might do better overall with a balanced blend of market, conversion, and value, rather than by extreme measures to absolutely maximize any one in particular.
In application, Adam argues that database companies, especially those using new license terms, like Mongo and Elastic, focus too much on making sure they’re the only ones who can monetize the open source projects they steward. That prevents them growing addressable market size as they otherwise might. To mix banal business metaphors, they jealously guard all the pie for themselves, when they could be sharing slices and eating more overall.
The zinger: From this point of view, seeing an AWS or a GCP or an Azure launch a hosted version of your open source offering should come as a welcome sign. It creates more awareness of the product and entrenches it more firmly as an industry standard, expanding your addressable market. Sure, your new competition will take some leads that might have been yours. But if the creator of the software can’t differentiate itself from generic, relatively lazy followers-on, something has gone very, very wrong.
The natural retort, which didn’t come up on the podcast, asks why the advantages that make an AWS so much better than you at expanding addressable market can’t also be used to beat you on conversion. Perhaps Mongo or Elastic or the next vendor can offer substantial additional features, services, assurances, and so on. But is that enough to overwhelm the network effects of AWS name recognition, marketing budget, sales presence, integration, financial girth, a contract likely already in place, and all the rest?
It’s common cliche that an upstart has to offer a ten-times value improvement to unseat an incumbent entrenched by network effects. The number isn’t so important as the gist: your upgrade has to be a lot, lot better, functioning more as the start of a new category than an improvement on the old one.
It’s one thing to say the firm stewarding an open source project should be able to differentiate itself, as if the playing field were level. On a level field, having the leader of the open project might be enough. But it’s quite another to say that the firm ought to be able to demonstrate overwhelming superiority, especially when business, transaction-cost, and financial factors start to overwhelm technical details in procurement. You have to be overwhelmingly better as a product, but also, in many ways, at least comparable as a business. Especially if you target slow-lead, large-dollar, long-term enterprise deals.
One way to clear such a high bar is to make a ton of new value, all the time, at a relentless pace. The Business Source License and other time-delay release pacts implement exactly that kind of commitment in legal terms. Business-wise, it’s a treadmill. Stand still, you fall off the back.
All that sounds nice in the abstract. Innovation is supposedly why we tolerate all the heat and noise of competition to begin with. But keeping unbroken pace is particularly hard with typical turnover and burnout time frames common at VC-backed companies. More so since they tick down first for key founders and early hires, often the leading names of the open project that provide obvious differentiation. In sum, the treadmill approach effectively turns software plays into professional services plays. You’re not getting paid for what you’ve done, but what you will do. People, they say, do not scale.
Another approach, complimentary rather than exclusive, is to withhold value created out of the open product, so it’s not part of what your new, entrenched competitors get for free. Shrink your open core, and allocate more of the value you deliver to the closed shell of add-ons around it. Which is exactly what we’re seeing with stronger copyleft licenses, as from Mongo, source-available add-ons, as from Cockroach, and features exclusive to the steward company’s hosted offering, as from Elastic.
Adam is doing some soul searching about the meaning of free and open source software. What is this thing that he’s identified with, promoted, and benefited from for so long?
I have done a bit of that myself. But as engrossing as the personal process may be, for proper perspective, I think we have to remember: If we have time for an identity-meaning-purpose crisis, that’s a pretty sure sign we feel our more basic needs securely met. Securing more basic needs is most folks’ day-by-day. “What did it all mean?” is winner’s mystery.
I don’t think that everything that binds a software project together ought to end up in a license. But I do believe in writing things out and passing them around. The process teaches us what we actually agree on. It’s a form of discipline, a reliable antidote to nod-along consensus that achieves detente by avoiding disagreement, each member projecting their own views onto all the others. Then, eventually, some latent dust-up comes to the fore, compounded by the element of surprise. Open Source is rife with this powder-keg potential, like Free Software before it.
If the gospel of open source boils down to nothing more than good social feelings from giving people like us what they want, I’m not surprised we come up feeling a penny short of transcendence. When the good we do ourselves—ripe social jollies—is so concrete, but the good we do for others—even others like us, to say nothing of society more broadly—so nonspecific, we are at serious risk of self-service in disguise, of edifying our own insularity. When we’re doing what feels good, going back to make it sound meaningful isn’t hard. We only have to convince ourselves.
If I heard policemen or accountants waxing philosophical about how camaraderie among peers is all that matters, I’d cry foul. I’d expect others to recoil from the same line out of a lawyer. Ideally, everyone who practiced law would enjoy it. We’d end up with better lawyering that way. But lawyering for lawyering’s sake? There lie the old seeds of corporatism, of conspiracy against the laity. In the end, people program computers for a reason. If we seek a lasting sense of greater purpose and achievement, we’ll found it outside our guild, not burrowing deeper within it.
Some of the folks getting what they want from open source these days are profiling grandparents for advertisers, hocking kleptotokens to the technically illiterate, feeding AI automation woo-woo to managers fighting union drives, surveilling the interesting kids we knew at college for the government, and colluding to suppress developer salaries, in part by prolonging a naivete about pay, spec work, and reputation that would be shocking in any other industry. Developers doing all of those things are getting what they want from open source.
That’s the writing on the wall. I can’t pretend not to read it. I’m not self-actualized when I do.
Your thoughts and feedback are always welcome by e-mail.
more articles — back to top — edit on GitHub — revision history