Written advice: Difference between revisions

From The Jolly Contrarian
Jump to navigation Jump to search
No edit summary
Tags: Mobile edit Mobile web edit Advanced mobile edit
No edit summary
 
(4 intermediate revisions by the same user not shown)
Line 1: Line 1:
{{essay|systems|written advice|}}{{drop|W|e live in}} a world increasingly comprised of what [[Lorraine Daston]]<ref>{{br|Rules: A Short History of What We Live By}}</ref> would call “thin” rules — specific, formalistic, algorithmic micro-rules designed for literal and measurable compliance. Contrast these with “thick rules”, which are generalised principles requiring comprehension, judgment and balance. “avoid [[conflicts of interest]]”. “[[Client’s best interest rule|Treat your customers fairly]]”.  
{{essay|systems|written advice|}}====Algorithms and heuristics====
{{drop|A|s we lie}} back and let the machines roll over us and the village square digitises, the relatively few ''principles'' by which we live give way to a multitude of ''rules''.<ref>[[Lorraine Daston]], in her excellent {{br|Rules: A Short History of What We Live By}}, calls these “thin” rules — specific, formal, algorithmic micro-rules designed for literal and measurable compliance without tolerance. Principles she calls “thick rules” general [[heuristic]]s requiring comprehension, judgment and balance: “avoid [[conflicts of interest]]”. “[[Client’s best interest rule|Treat your customers fairly]]”.</ref> The problem is this: [[Turing machine|our machine overlords]] are good at following rules but bad at understanding principles. Humans are good at understanding principles but bad at following rules. When you assign machines, and ''processes'' to make decisions and humans to push buttons you have badly bished the [[division of labour]].


It is part of JC’s ongoing mantra that as we surrender ourselves to the [[deterministic]] dogma of [[Scale|scale]] wherein we can and should be managed by [[algorithm]] — where we trust the deterministic predictability of [[algorithm]]s and distrust the ineffable, unjudgeable [[metis]] and human experience — not measurably “fair” in the sense of even and predictable — we lose something very important.  
Entrusting our future to deterministic [[algorithm]]s over our own [[metis|expertise, empathy and experience]] is, we think, an unseemly surrender for team ''homo sapiens''. For [[thick rules|principles]] do not reduce themselves conveniently into [[thin rules|algorithms]]: literature never tires of reminding us of this: this is the moral of the “be careful what you wish for” genre: Aladdin and his lamp, the sorcerer’s apprentice, Macbeth: whenever a protagonist cuts a corner, stops concentrating, or consults a fortune teller.  Models behave badly. We see “[[Normal distribution|25 standard deviation moves, several days in a row]]”. The truth is complicated.


Anyhow.
Most technology aims to short-cut graft and goose scale to reach a desired end. We bewitch broomsticks. Nowhere more so than in regulation, where principles of conduct — requiring expertise, judgment and trust, both from regulator and regulated, have given way to detailed rules, intended to need neither.


The problem is that [[thick rules]] do not reduce themselves conveniently into [[thin rules]], as our literature never tires of reminding us: this is the essential moral of any story in the “''be careful what you wish for''” genre: Aladdin and his lamp, the sorcerer’s apprentice; anything in which the protagonist takes a lazy shortcut to achieve a comfortable end.
You can generalise averages, medians and trends from specifics well enough. But you cannot reverse-engineer specifics from the general. [[Emergence|Emergent]] properties are odd things: you gain information that isn’t in the constituent set; you lose information that is.  


While they are meant to be deterministic and clear, the practical [[thin rules]] will inevitably be dispersed across various sources: legislation, promulgated regulations, previously decided cases, public guidance, common law principles, and analogy to the European regulation from which the rule was originally derived and in some case woodenly translated.  
We [[Robomorphism|robomorphise]] at our peril.
 
In any case, the job is increasingly to comply with detailed, strict, narrow rules. We need not — indeed ''should'' not — consider wider principles that the rules are meant to deliver: that is, literally, above our pay grade.
 
“Excellence” in the art of thin rule formulation is clarity: a given rule must be clear, simple, actionable, and free from doubt for someone with basic reading comprehension.
 
“Thin rules” are like atomised bits of code: running the code is a matter of [[symbol processing]] not [[Interpretation and construction|literary construction]].
 
But in practice [[thin rules]] will inevitably be dispersed across various sources: legislation, promulgated regulations, previously decided cases, public guidance, common law principles, and analogy to the European regulation from which the rule was originally derived and in some case woodenly translated.  


Now, nowhere is more beset with thin fiddly rules than the financial services industry. The JC has [[The domestication of law|a theory]] that the financial services industry has evolved as a means of interpreting rules, and not ''vice versa''. There is some irony that those who make the rules are generally not prepared to say what they mean, so if one wants to know, one must seek the, well [[metis]], of one who holds themselves out as an expert and who is, for a fee, prepared to say.  A lawyer. Having a kind solicitor sort all that out and some it up in a memo is a price worth paying!  
Now, nowhere is more beset with thin fiddly rules than the financial services industry. The JC has [[The domestication of law|a theory]] that the financial services industry has evolved as a means of interpreting rules, and not ''vice versa''. There is some irony that those who make the rules are generally not prepared to say what they mean, so if one wants to know, one must seek the, well [[metis]], of one who holds themselves out as an expert and who is, for a fee, prepared to say.  A lawyer. Having a kind solicitor sort all that out and some it up in a memo is a price worth paying!  
Line 24: Line 33:
   
   
Now, [[private practice lawyer]]s, listen up. How you respond to this question depends on what you think the answer is. But  imagine you are a dentist welcoming a new patient to your surgery.  
Now, [[private practice lawyer]]s, listen up. How you respond to this question depends on what you think the answer is. But  imagine you are a dentist welcoming a new patient to your surgery.  
====You are sure you can===
====You are sure you can====
Let us say you are confident there is an easy answer, and it is, “yes, you may do Y”.
{{drop|L|et us say}} you are confident there is an easy answer, and it is, “yes, you may do Y”.


First thing: ''tell your client this''. Orally: let them know you can give them the answer they want, and in a short time frame and for a sensible price, you will prepare a memorandum of advice that they can use for internal approvals, arse covering and plausible deniability — the sacred quest of the inhouse lawyer.  
First thing: ''tell your client this''. Orally: let them know you can give them the answer they want, and in a short time frame and for a sensible price, you will prepare a memorandum of advice that they can use for internal approvals, arse covering and plausible deniability — the sacred quest of the inhouse lawyer.  
Line 34: Line 43:


====You are sure you cannot====
====You are sure you cannot====
If you are sure the easy answer is, “no, you may not do Y”. Then tell the client, explain your reasoning orally, and ''then put the phone down and close the file''. You might make a telephone attendance note recording your advice for the file — this is to cover ''your'' arse, should the client subsequently dispute it — but ''you should not put your negative advice in writing''. The simple reason for this is that it every time you put pen to paper it costs your client money, and no-one wants to pay extra for written confirmation of bad news you have already given them.  
{{drop|I|f you are}} sure the easy answer is, “no, you may not do Y”. Then ''get on the phone''. tell the client, explain your reasoning ''orally'', and then ''put the phone down and close the file''. You might make a file note recording your advice to cover ''your'' arse, should the client subsequently dispute it — but unless the client asks you to, ''do not send it to the client''. Generally, one should not put negative advice in writing. The simple reason for this is that it every time you put pen to paper it costs your client money, and no-one wants to pay extra for bad news you have already given them.  


But there is a deeper epistemological reason — this may be controversial in light of the [[Post Office Horizon IT scandal|Post Office farago]] but it is still true — and that is that it preserves the ''formal'' possibility of that action. There may be nuances, edge cases, extrapolations and contexts where action Y is, for subtle reasons permissible. ''You may be wrong''. Once that advice is on the record, your client has a formal problem in acting contrary to that advice even if, substantively, that action would be justified in the circumstances. Do not put your client on written notice of things it has not asked you to give it notice about.
But there is a deeper [[epistemological]] reason — this may be controversial in light of the [[Post Office Horizon IT scandal|Post Office farago]] but it is still true — and that is a written record of bad news does not help your client, and may make things worse. Not committing to bad news to writing preserves the ''formal'' possibility of that action. There may be nuances, edge cases, extrapolations and contexts where done variation on the action is permissible. ''You may be wrong''. Once your negative advice is on the record, you have, without helping your client, given it a formal problem. If it then acts contrary to your advice even if, substantively, it is justified in doing so — she has assumed personal risk ''in the face of contrary advice. Do not ask your client to pay you to formally put her on written notice of what she should not do. ''Unless she asks you to''.  


Legal memoranda are like Christmas letters from distant aunts: confections of happy news, telling you how well young Fortescue is doing with his water polo, the lovely summer holiday all had at Bognor Regis, the lovely kitchen renovations, but no word about Neville’s failing marriage, Imogen’s kleptomania or Emilia’s out of control coke habit. ''Spare the bad news. The whole family knows about that already.''
Legal memoranda are like Christmas letters from distant aunts: confections of happy news, telling you how well young Fortescue is doing with his water polo, the lovely summer holiday all had at Bognor Regis, the lovely kitchen renovations, how grateful we should all be for circumstances, however outwardly meagre — but there will be no word of Neville’s failing marriage, Imogen’s kleptomania or Emilia’s out of control crack habit. ''Spare the bad news. The whole family knows about that already.''


Remember your client’s overarching principle of “plausible deniability”.  
Remember your client’s overarching principle of “plausible deniability”.  


====You think you can, but are not sure====
====You think you can, but are not sure====
This is where you earn your keep. You get paid for your legal research. Ideally this would be a novel or subtle question — if it isn’t you should know it, and should not be charging a client for finding out if you don’t — but the sequence is: orally, “this sounds okay but there are subtleties, let me quickly check those out and come back to you” — whereupon do your stuff, reconvene,  
{{drop|T|his is where}} you earn your keep. You get paid for your legal research. Ideally this would be a novel or subtle question — if it isn’t — if a competitor would know this of the top of her head, should you be charging your client for catching up? — but the sequence is: orally, “this sounds okay but there are subtleties, let me quickly check those out and come back to you” — whereupon do your stuff, reconvene, and write a nice short, clear memorial. This helps your client — give her wings. If there are qualifications or caveats it is quite all right to be self-serving about them: rather than:
 
{{Quote|
“It is quite all right to to Y, however in no circumstances should you do Z”}}
 
Consider:
 
{{Quote|
“This advice considers Y; we have not addressed Z. If you have any intention in that regard we would be happy to consider it for you.”}}


{{Sa}}
{{Sa}}

Latest revision as of 10:35, 21 June 2024

The JC’s amateur guide to systems theory
Index: Click to expand:
Tell me more
Sign up for our newsletter — or just get in touch: for ½ a weekly 🍺 you get to consult JC. Ask about it here.

Template:M intro systems written advice

Premium content
Here the free bit runs out. Subscribers click 👉 here. New readers sign up 👉 here and, for ½ a weekly 🍺 go full ninja about all these juicy topics👇

See also

Template:M sa systems written advice

References

Algorithms and heuristics

As we lie back and let the machines roll over us and the village square digitises, the relatively few principles by which we live give way to a multitude of rules.[1] The problem is this: our machine overlords are good at following rules but bad at understanding principles. Humans are good at understanding principles but bad at following rules. When you assign machines, and processes to make decisions and humans to push buttons you have badly bished the division of labour.

Entrusting our future to deterministic algorithms over our own expertise, empathy and experience is, we think, an unseemly surrender for team homo sapiens. For principles do not reduce themselves conveniently into algorithms: literature never tires of reminding us of this: this is the moral of the “be careful what you wish for” genre: Aladdin and his lamp, the sorcerer’s apprentice, Macbeth: whenever a protagonist cuts a corner, stops concentrating, or consults a fortune teller. Models behave badly. We see “25 standard deviation moves, several days in a row”. The truth is complicated.

Most technology aims to short-cut graft and goose scale to reach a desired end. We bewitch broomsticks. Nowhere more so than in regulation, where principles of conduct — requiring expertise, judgment and trust, both from regulator and regulated, have given way to detailed rules, intended to need neither.

You can generalise averages, medians and trends from specifics well enough. But you cannot reverse-engineer specifics from the general. Emergent properties are odd things: you gain information that isn’t in the constituent set; you lose information that is.

We robomorphise at our peril.

In any case, the job is increasingly to comply with detailed, strict, narrow rules. We need not — indeed should not — consider wider principles that the rules are meant to deliver: that is, literally, above our pay grade.

“Excellence” in the art of thin rule formulation is clarity: a given rule must be clear, simple, actionable, and free from doubt for someone with basic reading comprehension.

“Thin rules” are like atomised bits of code: running the code is a matter of symbol processing not literary construction.

But in practice thin rules will inevitably be dispersed across various sources: legislation, promulgated regulations, previously decided cases, public guidance, common law principles, and analogy to the European regulation from which the rule was originally derived and in some case woodenly translated.

Now, nowhere is more beset with thin fiddly rules than the financial services industry. The JC has a theory that the financial services industry has evolved as a means of interpreting rules, and not vice versa. There is some irony that those who make the rules are generally not prepared to say what they mean, so if one wants to know, one must seek the, well metis, of one who holds themselves out as an expert and who is, for a fee, prepared to say. A lawyer. Having a kind solicitor sort all that out and some it up in a memo is a price worth paying!

Now, no thoughtful client brings such a question to a lawyer without also having in mind a preferred answer to it. Something along the lines of “it is fine for me to do/not do this,” depending on the commercial implication.

The client may not be realistic in that aspiration. An in-house lawyer may know perfectly well it is a fruitless exercise but will do it to get her salesperson off her back. This is a poor use of legal counsel, but we should not fool ourselves it does not happen. But most of the time there is room for doubt.

For good commercial lawyer, there are obvious rules of engagement here — but the world is beset with poor commercial lawyers.

The first is this: bear your client’s desired outcome in mind. If you do not know it, ask.

The proposition is usually:

“I wish to be able do Y. There is a new Regulation X on the topic of Y. To our mind, Regulation X is ambiguous. It could be interpreted to mean A or B. If it is A, then we can do Y. If it is B, then we cannot. What is your advice about Regulation X? Can we do Y?”

Now, private practice lawyers, listen up. How you respond to this question depends on what you think the answer is. But imagine you are a dentist welcoming a new patient to your surgery.

You are sure you can

Let us say you are confident there is an easy answer, and it is, “yes, you may do Y”.

First thing: tell your client this. Orally: let them know you can give them the answer they want, and in a short time frame and for a sensible price, you will prepare a memorandum of advice that they can use for internal approvals, arse covering and plausible deniability — the sacred quest of the inhouse lawyer.

Lawyer can then go away, greenlight the business to get cracking, and your memo can follow in due course.

Only do this if you are really confident you can give a clean answer. There is nothing worse than legal advice that does not come up to brief.

You are sure you cannot

If you are sure the easy answer is, “no, you may not do Y”. Then get on the phone. tell the client, explain your reasoning orally, and then put the phone down and close the file. You might make a file note recording your advice to cover your arse, should the client subsequently dispute it — but unless the client asks you to, do not send it to the client. Generally, one should not put negative advice in writing. The simple reason for this is that it every time you put pen to paper it costs your client money, and no-one wants to pay extra for bad news you have already given them.

But there is a deeper epistemological reason — this may be controversial in light of the Post Office farago but it is still true — and that is a written record of bad news does not help your client, and may make things worse. Not committing to bad news to writing preserves the formal possibility of that action. There may be nuances, edge cases, extrapolations and contexts where done variation on the action is permissible. You may be wrong. Once your negative advice is on the record, you have, without helping your client, given it a formal problem. If it then acts contrary to your advice — even if, substantively, it is justified in doing so — she has assumed personal risk in the face of contrary advice. Do not ask your client to pay you to formally put her on written notice of what she should not do. Unless she asks you to.

Legal memoranda are like Christmas letters from distant aunts: confections of happy news, telling you how well young Fortescue is doing with his water polo, the lovely summer holiday all had at Bognor Regis, the lovely kitchen renovations, how grateful we should all be for circumstances, however outwardly meagre — but there will be no word of Neville’s failing marriage, Imogen’s kleptomania or Emilia’s out of control crack habit. Spare the bad news. The whole family knows about that already.

Remember your client’s overarching principle of “plausible deniability”.

You think you can, but are not sure

This is where you earn your keep. You get paid for your legal research. Ideally this would be a novel or subtle question — if it isn’t — if a competitor would know this of the top of her head, should you be charging your client for catching up? — but the sequence is: orally, “this sounds okay but there are subtleties, let me quickly check those out and come back to you” — whereupon do your stuff, reconvene, and write a nice short, clear memorial. This helps your client — give her wings. If there are qualifications or caveats it is quite all right to be self-serving about them: rather than:

“It is quite all right to to Y, however in no circumstances should you do Z”

Consider:

“This advice considers Y; we have not addressed Z. If you have any intention in that regard we would be happy to consider it for you.”

See also

References

  1. Lorraine Daston, in her excellent Rules: A Short History of What We Live By, calls these “thin” rules — specific, formal, algorithmic micro-rules designed for literal and measurable compliance without tolerance. Principles she calls “thick rules” general heuristics requiring comprehension, judgment and balance: “avoid conflicts of interest”. “Treat your customers fairly”.