|The Devil’s Advocate™
The reg tech proposition is obvious: replace this laborious, error-prone, analogue, human process with an authenticated, governed, audited, straight-through processed, fast online digital interaction. So why doesn’t it work, and what can we do about it?
- It’s iatrogenic: Because the provider’s primary interest is its annuity, the cure tends, in practice, to be worse than the disease.
- It’s expensive: Every participant is in it to make a turn. Their interest is in participating, and making a turn first, and only then in the “desired outcome”.
- It’s prone to use-case obsolescence: Its proprietary nature means reg tech tends to be tightly controlled, top-down managed and targeted abstractly at a perceived demand and an anticipated future state. But the future, as imagined by the thought leaders of the reg tech, was forged in the now, which tomorrow will be the past. Just as previously imagined futures from the past have proven — flying cars, off-world replicants, the colonisation of Mars, the singularity, a world without work — predicting the future was, and remains, hard. The answer is not to try: to leave the architecture open to users to imagine as they go. No-one predicted SETI@home, after all. But “leaving everything to the user” doesn’t give a rentier capitalist much to do, so rent-seekers tend to constrain their products, requiring paid-for development, and consigning them to short-term obsolescence without continual maintenance. And maintenance means rent.
- It’s prone to competition obsolescence: To treat what should be a utility as a profit opportunity exposes you to another source of obsolescence. Competition. How can you know your platform will be the platform? How do you keep that position once you’ve got it? Friends Reunited ring a bell?
We’re yet to find a reg tech provider whose business model does not involve extracting rent. This they commonly justify by reference to the value their product provides, which they equate to the total cost of labour and infrastructure it saves. Historians, and those who enjoy irony, will notice how, in equating the value of a service to the net amount of labour it requires — or in this case “saves”, this resembles the labour theory of value.
You might ask how a vendor could know. You might also wonder what there is to gain by cutting the wage bill of some school-leavers in Bucharest, if you pay most of what you save away again, to a start-up in Old Street which bought some code from some school-leavers in Bucharest. Especially if that vendor will then be able to intermediate your process for the hereafter, adding the cheerful chime of a clipped ticket each time his machine spits out another document, or while your stuff collects dust on his server.
Wasn’t the promise of the information revolution grander than that? Weren’t things meant to be free, not just marginally cheaper but a lot more complicated?
Iatrogenic — it is a cure that is worse than the disease — in that in promising to alleviate the tedium of all that boilerplate — all the pernickitiness and the low-level wrangling over representations and warranties — technology throws open the window on a world of exactly that kind of low-level tinkering. The cost of pedantry has plummeted.
In the olden days, the anality of contracts was bounded by any lawyer’s natural capacity — deep, to be sure, but ultimately finite — to hold a superstructure of piecemeal salutary conditionality in her head. Nowadays, contracts can be infinitely, infinitesimally varied and endlessly customised. One can cater for any predilection, whim, doubt or proviso. We can put one in every line, if that is our wish. We can command that a space between words must be, or not be, italic.
This should not be a surprise. The word-processor did not shorten legal discourse. Email did not truncate communication. When Andy gaveth, it was not Bill, but our yen for verbal diarrhoea that tooketh away.
It won’t, if the reg tech rent-seekers have any say in it. Note the misalignment of interests: vendors have a direct incentive — in fact, a need — not just to fix, but to continue “fixing”, because “fixing” is how they get paid. They design their disintermediating machines to only disintermediate so far: users must remain dependent enough on their code, their systems and their expertise, that they have to pay an annuity for it. To pay rent.
The fault in our stars
Our friends in the management consulting profession (also rent-seekers, needless to say) encourage this disposition through the dogma of outsourcing. Here the gist is: if you have a convoluted process that is costing you time and money, outsource it, to someone better specialised, incentivised and remunerated to do it, who can do it cheaper, better and — thanks the magic of Adam Smith’s invisible hand — at the optimal cost. In this way do we entrench rent-seekers, by building an entire (rent-seeking) infrastructure around this newly articulated process — with its own middle management, operations, compliance, internal audit, procurement, you name it — yes, even legal — without ever asking whether the process was all that important in the first place.
Once imposed, the new machinery is so costly to remove that it is easier just to live with it, from time to time tweaking it, optimising it, adjusting it, relocating it from a service centre in Bangalore to one in Manilla, every little fiddle keeping that entire bureaucratic iron lung fully engaged. And no management consultant will ever tell you the obvious truth: had you decomplicated the process in the first place — or, hang it, just got rid of it — at the cost of some some peripheral academic protections, none of this would have been necessary.
But, but, but! The risk! The organisation must be protected!
So where does this leave us?
The cross default challenge
Here is a challenge: one day, out of the blue, ask a risk officer to explain, exactly, what cross default is and how it works. On the spot. No phone-a-friend. No ask-the-audience. Expect blank looks and notebooks.
Now ask yourself this: it it better to keep that peripheral right to cross default with all the tedious arguments it invites — whether triggered by default or acceleration, as to the level and symmetry of thresholds, what counts as indebtedness and whether to include derivatives or deposits, what sorts of grace periods should be allowed, whether to carve out operational errors or settlement failures — is it better to keep all of that fidgetery and move the people who handle it out to Hanoi — or just strike it out of your contract forms altogether?
As long as cross default seems like a competitive advantage; as long as we think our boilerplate is some kind of proprietary technology, we consign ourselves to a permanent arms race with our own shadows.
Satisfaction is unsatisfactory
Mick Jagger and Keith Richards composed Satisfaction in 15 minutes whilst sitting by a pool in Florida. They are still, 55 years later, reaping colossal dividends from that inspired quarter hour. Fair play to them: this may be the law, but it is not good law. It is a bad model for reg tech.
Rent-seeking encourages us to squirrel away verbal mush on the grounds that it is “proprietary technology”. There is nothing — utterly nil — that is special, clever or even desirable about legal boilerplate.
Boilerplate is the point where marginal utility touches zero. That ISDA schedule — yes, friends, the whole god-forsaken thing, even the customised ATEs? No-one cares. No-one likes it. No-one wants it. It is a regrettable externality; a grim fact of life. Legal negotiation is the dreary formality one must go through to get to the heart of the deal. It is not the sausages; it is the mask you need to put on so you can go into the supermarket and get the sausages.
Extended phenotypes and the mercurial arrow of causation
So these feedback loops drive us away from the place we want to go. Instead of a self-service ethos of simplified, standardised, user-friendly components that, though continual refinement, are honed, standardised and simplified, we have processes which keep their existing convolution, get worse, all the time supporting a more elaborate infrastructure that purportedly is there to support them.
Recall once again, Yuval Harari’s observation about the domestication of wheat, and apply it here: is your reg tech implementation, and its attendant support structure, some kind of extended phenotype that has evolved: a marvellous beaver dam; an adaptation wrought magically on the environment so this delicate but vital process can survive?
Which way does the causal arrow flow? Who domesticated whom?
Staying with this process is not a zero-cost option. This infrastructure is monstrously expensive. Every year, it costs tens or hundreds of millions of dollars. For each firm. Contract negotiation is not a profit centre for anyone but its practitioners.
Spend good money here, instead of throwing it after bad money you’ve spent there. Design an open source, open-architecture system with feedback loops that push us towards a simplified, standardised centre, rather than away from it. Emphasise transparency and commonality, not proprietary technology. In essence, build a GitHub for legal terms.
See ClauseHub for a high-level technical articulation of the platform.
- Service level agreement
- Management consultant
- Subject matter expert
- ClauseHub: theory
- Legal services delivery
- Why is reg tech so disappointing?
- Thought leaders are no better at predicting the future of legal services than they have been at anything else.
- The internet was revolutionary because it imagined no future, but left that — and continues to leave it, right? — to users. By the way, BOINC is a bit of a clue to where this is all going.
- This is called “hosting” and it seems to be a cash cow. But aren’t terabytes of data storage, like, pennies these days?
- This is a super concept and if you haven’t come across it you owe it to yourself and Nassim Nicholas Taleb to read about it in his superbly bombastic Incerto series.
- When the great J. M. F. Biggs first isolated and documented a Biggs hoson in the wild — the celebrated bold full stop in a “Boats” repackaging — it was a once-in-a-generation event.
- I have lost count of the times that that a tech provider has told me users cannot have edit or configuration rights on a piece of software. There are two explanations for this, and neither is edifying: one is that the software is so fragile or poorly designed that allowing a user to tinker with it will make it break; another is that it is so basic that allowing users to see it will reveal how simple it really is.
- Listening, as they do, to tracks by Sam Therapy and King Dice...
- I confess this is a bold assertion and I may be over my skis on it, but I have done some research. I am yet to find any example of a crust fault being exercised in an isda. If you know of one, please do get in touch. Confidentiality and discretion is assured: I just would like to know.
- I take a liberal view of boilerplate as “anything that doesn't make it into the termsheet”.
- And no, the “user” is not the legal eagle: the user is the client.
- Harari’s suggestion, which owes something to Douglas Adams and his pan-dimensional hyper-intelligent mice, is that wheat domesticated homo sapiens and not vice versa. See https://www.ynharari.com/topic/ecology/