I’ve been reading a lot about AI recently. Can’t be avoided, although some days I try. Everyone in the chattering class has strongly held views, albeit, I suspect the punditry really knows no more about AI than the rest of us. All I want to know is if I should liquidate my Christmas club account now as I won’t make it to Christmas to spend it. I recognize that “we’ll all be defenestrated by AI before Christmas” is a minority view. I’m going to dismiss it.
Apparently, “guardrails” will protect us. Guardrails are all over our conversations about AI. Guardrails, we are told, are the things that keep artificial intelligence systems operating safely. Guardrails will keep us from having a Matrixmoment. Guardrails will prevent AI from going all SPECTRE on us. Guardrails will make sure that I don’t have to liquidate my Christmas Club before Christmas…at least this year.
What really are guardrails? A little research illuminated that the word guardrails was first used in the middle of the 19th century to describe physical barriers that were built around railroad tracks. It was a sensible notion because wandering in front of a moving train is…sub-optimal and, at that time, the folks perhaps weren’t always clear about what a railroad really was and how fast it might move.
In the 21st century the phrase has been transmogrified to convey the notion that there are limitations on people’s behavior. That’s noteworthy because the original meaning of guardrails was to keep something out, while the new use is to keep something in. That change seems to me to be meaningful. Now, guardrails is often used to describe the outside boundaries of acceptable behavior; behavior for which one would not expect to be censured for violating social conventions.
This is what I want to talk about. Think of guardrails as squishy societal standards that are adjacent to norms, but are considerably less constraining. If you can’t expect people to do what they’re actually supposed to do (traditionally captured in the notion of norms), perhaps we can at least establish boundaries to keep them from being complete shits. Guardrails are broad while norms are narrow and strict. Have we given up our norms to embrace these less constraining fuzzy guardrails? Indeed, in many cases, we have and that’s a problem.
To be a bit over the top here, consider homicide. A truly central norm of civilized life is don’t commit murder. That’s pretty straightforward. Are we now migrating towards a guardrail where murder can be excused based on the context of the doing? Is our new guardrail “don’t commit murder unless a meaningful segment of the population thinks the poor bastard deserved it? (And if you whack someone today the folks don’t like, you might even get a GoFundMe account and maybe a couple of endorsement contracts for, I don’t know, face products, hand guns and the like.)
Norms allow you to proceed through an intersection when the light is green. If traffic lights become guardrails that merely suggest you shouldn’t run into anyone in an intersection if you can possibly help it, everyone would need to creep through intersections to make sure someone wasn’t testing the outer boundaries of the new intersection guardrails.
Guardrail use diminishes the importance and even the significance of norms. It makes our behavior less predictable…generally, in a rather bad way.
Guardrails excuse bad behavior which, at one point, would have been considered unacceptable . Now, such behavior might be seen as just sort of icky, you know, not great but not entirely beyond the pale. You can behave within the outside boundaries of guardrails, far from the norms, and still show up at the country club the next day without fear of embarrassment or censure.
Guardrails degrade our expectation of what people will or will not do. If we can’t expect people to meet long-established societal norms, everything gets harder. By necessity, we daily make assumptions on how the people we’re dealing with will act.
In business, the substitution of guardrails for norms means that a lot of bad behavior is tolerated, indeed maybe almost expected. Are blacklines accurate? (Oops, I took this blackline off a prior draft…sorry!) Are verbal agreements likely to be respected? Might truthful-ish be substituted for truth? Might foot faults be seized upon for advantage? Can I fool my counterparty into making a mistake with an intentionally incomprehensible and turgid word salad? Can I intentionally obscure meaning for advantage?
Norms are shorthand to facilitate interpersonal engagement. The shorthand allows us to assume a lot about our counterparty’s behavior. These culture-wide cognitive shortcuts are the oil in the machinery of commerce and, in fact, in interpersonal relationships. When norms degrade towards wide guardrails in business, we need to fabricate detailed, specific, prescriptive rules, to replace and mimic those missing norms. It’s the substitution of prescription for principle and it’s exhausting.
Since I largely live in the world of commercial real estate finance, I can easily find examples from that space. Think of the servicing standard in CMBS. When it was originally conceived, it meant I’ll service the loans with the same care and attention that I’d service my own loans and without regard any conflicts of interest I might have in doing so. That’s an easy to understand 30 words. While that was the norm, and you expected the norm to be met, perhaps that was all we needed. Now, we employ several hundred words resulting in often opaque and largely unreadable dense text to get to the same place. The same place! What have we accomplished? Documents we use every day in our professional lives, things such as indentures, servicing agreements, mortgages, assignments, leases, participation agreements, intercreditor agreements, etc., etc., have bloated from relative brevity at inception to hundreds and hundreds of pages (and, by the way, continuing to grow). This makes everything harder. In fact, all this spaghetti-bowl text can, even if well intended, become fertile grounds for mistake. It also makes everything more expensive. It makes all negotiations take much longer. It distracts counterparties from focusing on the main deal feature while being sucked down a rabbit hole of often nonsensical minutia. Only the lawyers can be happy.
While perhaps expecting norms always to be met was a bit aspirational, at least it was an anchor and now we’ve lost that anchor.
We theoretically assume that investors in private placements , institutional accredited investors, are smart and understand the business in which they’re investing, but we don’t really act like they are, do we? If we did, would we feel compelled to lard our disclosure documents with 100 pages of risk factors, including such howlers as “investing in real estate involves risk.” They’re there because our guardrails allow, as acceptable behavior, a sophisticated investor to sue for failure to include a material fact, even if the investor knew full well they are not actually being misled and indeed were aware that the facts in question were broadly known and understood. So, investing in real estate can be risky, huh? Who knew? This gives rise to nonsensical risk factors, pages and pages (and dollars and dollars) of blindingly obvious drivel.
If you must assume people will behave badly, yet still be able to show up to the country club for breakfast the next morning because norms have been eroded and replaced with broad guardrails, you’d better make sure there’s no play in your agreements, in the meanings of your text; no ambiguity in the documentation whatsoever. Ambiguity cannot comfortably be let lie in a world where gotcha behavior is fine, indeed to be expected.
Haven’t you often thought that it would be delightful if could simple close on the term sheet? That’s where all the real issues are addressed. We all, at that point, know what the deal is. The real issues have been hashed out and everyone understands the what and why of it. So, why do we then spend weeks and weeks of tedious drafting and negotiation to replicate the same fundamental business deal? It’s because we don’t trust counterparties to do the right thing, unless we constrain them hand and foot with adamantine and unyielding, prescriptive rules.
My guess is that the genie can’t be put back in the bottle. Things lost can’t be restored. In our business transactions (and life, I guess), the only thing to do is look very hard for evidence that a counterparty is likely to ride those guardrails; that that counterparty is likely to test the outer edges of acceptable behavior and is largely unrestrained by the still clearly understood but now often ignored norms. We shouldn’t do business with them. Bad, venal and untrustworthy managers and owners do more harm and have doomed more commercial real estate transactions than actual problems with underlying assets. Isn’t it always the way that those guardrail testers, those norm-violating (often charming) rogues always have the most fabulous projections, coolest and swankiest and most sizzling presentations and trumpet the most favorable prospects? Shun the bad guys, no matter how appealing the deal might look on paper. It won’t be worth it when crunched time comes; it will and you’ll lose.