4 ms·

Imagining legislative text as if it were software code, this is akin to discovering that an engineer has been running a regex across pull requests which replace

by jka 7y ago

Imagining legislative text as if it were software code, this is akin to discovering that an engineer has been running a regex across pull requests which replaces 'if (safety-condition-encountered)' with 'if (false)' -- just prior to merge/deploy. What would happen to that engineer and code review process in a well-run tech organization? And why should we not apply the same principles to legal changes -- which affect our real world environment -- as we do for software changes which affect our application environments?

bordercases 7y ago
Software systems are meant to be totally consistent, both for the sake of being able to execute deterministicly on dumb machines, and for reliable performance in relatively static or funneled environments. Legal systems operate in constantly evolving environments and in thick, untransparent cases where the truth of the matter is typically unknown and needs are conflicting. You might be able to do this kind of thing in limited settings of organizational design or corporate policy – but you will not be able to completely replace federal law with software.
- jka 7y ago
  Even in evolving environments, it's important to have terminology that describes what is acceptable and what is not in clear terms. Writing precise language is hard, but ambiguity of language is worse, since it can lead to potentially unsafe interpretation. The Frankenstein Veto makes it possible for state Governors to alter the meaning 'near runtime' - certainly post-review - and accidentally or intentionally create loopholes which enable damaging outcomes.
  - bordercases 7y ago
    But laws are already not written ambiguously. They may be written with technical jargon which may be too specific or euphemistic, but that's not quite the same thing. What you would be advocating for at that point might a common language for law that avoids euphemism and keeps the reading level at a certain level of complexity, but doing so also limits the richness of language that we can use to describe legal agreements in the first place. Furthermore any software system that could handle the Frankensetin Veto would have to be extremely competent in natural language and somehow do model checking on all the combinations of cases that natural language can cover. Not going to happen in polynomial time buddy. Not for all sorts of law anyway. The problem that the Frankenstein Veto case demonstrates doesn't necessarily have to be handled in software either. It's a problem with the workflow of law, not the systems implementation.
    - jka 7y ago
      I'm asking for simple language, accessible to a broad audience, which allows rules to be written so that most people applying them to a set of example situations would arrive at the same results. Beyond that, what I'm suggesting is that it could be helpful to gradually make it easier for the public to view, comment on, and search proposed and historic legislative changes. Collaboration, tracking down bugs and coaching/mentoring in open source software is greatly aided by tools like GitHub; it's worth having concerns about whether such a service itself should be proprietary, but I do think the value it creates is worth looking at. If something similar could be created for legal texts, I think it's possible that there could be similar benefits. There may be risks, too. My suggestion isn't that we should devise a software system to implement the rules of the Frankenstein Veto; instead my theory is that the time-to-detect and time-to-recover from a process problem like the Frankenstein Veto would both be shorter in the presence of an open, collaborative, transparent legislative repository.
- walshemj 7y ago
  No sorry legal /parliamentary systems do have to be consistent A parliamentary/ legal system cannot have a motion saying A and have the executive turn around exactly do B
  - bordercases 7y ago
    Read "Myth of the Rule of Law" by John Hasnas, who is a professor of law at Georgetown. https://archive.org/stream/THEMYTHOFTHERULEOFLAWByJohnHasnas47/THE%20MYTH%20OF%20THE%20RULE%20OF%20LAW%20by%20John%20Hasnas-47_djvu.txt https://archive.org/stream/THEMYTHOFTHERULEOFLAWByJohnHasnas... It is the nature of law to be a negotiation among parties, with all of the properties of negotation that this implies – personal interests, wavering agreement. Believing that law can carry an objective authority disguises this reality and leaving the nature of law unacknowledged makes regulation of that system more difficult. You might argue that making law more formal or explicit can help here, but laws are already very formal and explicit, and being formal and explicit doesn't guarantee total consistency, maybe paraconsistency at best unless you're dealing with logics much less rich than those we can embed in natural language.
    - walshemj 7y ago
      How does this apply to parliamentary procedure? my argument is you cant once a law has been agreed and suddenly say oh no I am going to rewrite the law on my personal say so.
      - bordercases 7y ago
        You can come up with multiple conflicting interpretations during a case that are only resolved through further qualification and cases. This is a part of procedure as much as "personal say so" is not.
kragen 7y ago
Are you saying the governor's boss should fire him, or if he refuses to do so, should be fired by the state's CEO? The problem with your suggestion is that Wisconsin is not in fact a dictatorship; there is no CEO with whom ultimate responsibility rests. Instead, the various constituencies who might otherwise be shooting each other in the streets have agreed to abide by a fixed set of rules, known as "laws". Clearly in this case the rules turned out to favor the governor more than was intended, but the option of simply ignoring them as a result has significant downsides.
- maxerickson 7y ago
  The role of CEO is a good analogy with governor. (There usually are layers of responsibility above both; the board for a CEO and the people for a governor) If you look at it in those terms, I think it's more likely that they mean that voters would be well served by setting aside other interests and reforming the legislative process further (the article discusses various reforms) so that the governor is not able to take actions that are effectively unilateral, and also by not re-electing (or perhaps in the extreme, recalling) the governor.
  - kragen 7y ago
    > The role of CEO is a good analogy with governor. No, a CEO is a dictator. In the US, the state governor is a significantly different role from a dictator.
    - _jal 7y ago
      I generally find that when people start drawing comparisons between government and corporations, they're probably doing to ignore at least one major difference in order to paint government in a bad light. Not always, but usually.
    - maxerickson 7y ago
      In most corporations, the CEO takes direction from and serves at the leisure of the board (and if there isn't a board, there will typically be some other mechanism where the owners exert control). They typically have a broad mandate (which I suppose is what you are getting at), but they don't literally have unfettered control of the company. (CEOs that own majority stakes in their companies get close to complete control, but that's not what you are arguing, that control flows from the ownership, not the executive position).
pjc50 7y ago
Politics isn't an engineering process, it's a knife fight carried out with pens.
- bordercases 7y ago
  "I saw him kill three men with a pencil. A fucking. Pencil."
dragonwriter 7y ago
> Imagining legislative text as if it were software code As a software developer with a political science degree and some legal education, you basically shouldn't do this. > this is akin to discovering that an engineer Legislators and executives are even less like engineers in intended function than legal text is like software code. Legislative text isn't software code.b ,
- CameronNemo 7y ago
  I'm also a swe with a polisci degree (but no legal education or qualifications). I think the comparison between policy and code is useful, but not perfect. Are there any specific areas that are different between them that we should pay mind to?
  - dragonwriter 7y ago
    There's a lot, but a couple key points: (1) Legal text is often (arguably, most often in the US system) intended to be reified in application and/or in subordinate enactments (e.g., regulation); not only is it not executable code, it's often not intended as a complete specification. (2) executive and legislative authorities (at least, in the US system) are not like members of a single team, but by-design adversarial power centers. (They are more like customers with conflicting demands than engineers on a team.)

by jka 7y ago

bordercases 7y ago

Software systems are meant to be totally consistent, both for the sake of being able to execute deterministicly on dumb machines, and for reliable performance in relatively static or funneled environments. Legal systems operate in constantly evolving environments and in thick, untransparent cases where the truth of the matter is typically unknown and needs are conflicting. You might be able to do this kind of thing in limited settings of organizational design or corporate policy – but you will not be able to completely replace federal law with software.

jka 7y ago
Even in evolving environments, it's important to have terminology that describes what is acceptable and what is not in clear terms. Writing precise language is hard, but ambiguity of language is worse, since it can lead to potentially unsafe interpretation. The Frankenstein Veto makes it possible for state Governors to alter the meaning 'near runtime' - certainly post-review - and accidentally or intentionally create loopholes which enable damaging outcomes.
- bordercases 7y ago
  But laws are already not written ambiguously. They may be written with technical jargon which may be too specific or euphemistic, but that's not quite the same thing. What you would be advocating for at that point might a common language for law that avoids euphemism and keeps the reading level at a certain level of complexity, but doing so also limits the richness of language that we can use to describe legal agreements in the first place. Furthermore any software system that could handle the Frankensetin Veto would have to be extremely competent in natural language and somehow do model checking on all the combinations of cases that natural language can cover. Not going to happen in polynomial time buddy. Not for all sorts of law anyway. The problem that the Frankenstein Veto case demonstrates doesn't necessarily have to be handled in software either. It's a problem with the workflow of law, not the systems implementation.
  - jka 7y ago
    I'm asking for simple language, accessible to a broad audience, which allows rules to be written so that most people applying them to a set of example situations would arrive at the same results. Beyond that, what I'm suggesting is that it could be helpful to gradually make it easier for the public to view, comment on, and search proposed and historic legislative changes. Collaboration, tracking down bugs and coaching/mentoring in open source software is greatly aided by tools like GitHub; it's worth having concerns about whether such a service itself should be proprietary, but I do think the value it creates is worth looking at. If something similar could be created for legal texts, I think it's possible that there could be similar benefits. There may be risks, too. My suggestion isn't that we should devise a software system to implement the rules of the Frankenstein Veto; instead my theory is that the time-to-detect and time-to-recover from a process problem like the Frankenstein Veto would both be shorter in the presence of an open, collaborative, transparent legislative repository.
walshemj 7y ago
No sorry legal /parliamentary systems do have to be consistent A parliamentary/ legal system cannot have a motion saying A and have the executive turn around exactly do B
- bordercases 7y ago
  Read "Myth of the Rule of Law" by John Hasnas, who is a professor of law at Georgetown. https://archive.org/stream/THEMYTHOFTHERULEOFLAWByJohnHasnas47/THE%20MYTH%20OF%20THE%20RULE%20OF%20LAW%20by%20John%20Hasnas-47_djvu.txt https://archive.org/stream/THEMYTHOFTHERULEOFLAWByJohnHasnas... It is the nature of law to be a negotiation among parties, with all of the properties of negotation that this implies – personal interests, wavering agreement. Believing that law can carry an objective authority disguises this reality and leaving the nature of law unacknowledged makes regulation of that system more difficult. You might argue that making law more formal or explicit can help here, but laws are already very formal and explicit, and being formal and explicit doesn't guarantee total consistency, maybe paraconsistency at best unless you're dealing with logics much less rich than those we can embed in natural language.
  - walshemj 7y ago
    How does this apply to parliamentary procedure? my argument is you cant once a law has been agreed and suddenly say oh no I am going to rewrite the law on my personal say so.
    - bordercases 7y ago
      You can come up with multiple conflicting interpretations during a case that are only resolved through further qualification and cases. This is a part of procedure as much as "personal say so" is not.

kragen 7y ago

Are you saying the governor's boss should fire him, or if he refuses to do so, should be fired by the state's CEO? The problem with your suggestion is that Wisconsin is not in fact a dictatorship; there is no CEO with whom ultimate responsibility rests. Instead, the various constituencies who might otherwise be shooting each other in the streets have agreed to abide by a fixed set of rules, known as "laws". Clearly in this case the rules turned out to favor the governor more than was intended, but the option of simply ignoring them as a result has significant downsides.

maxerickson 7y ago
The role of CEO is a good analogy with governor. (There usually are layers of responsibility above both; the board for a CEO and the people for a governor) If you look at it in those terms, I think it's more likely that they mean that voters would be well served by setting aside other interests and reforming the legislative process further (the article discusses various reforms) so that the governor is not able to take actions that are effectively unilateral, and also by not re-electing (or perhaps in the extreme, recalling) the governor.
- kragen 7y ago
  > The role of CEO is a good analogy with governor. No, a CEO is a dictator. In the US, the state governor is a significantly different role from a dictator.
  - _jal 7y ago
    I generally find that when people start drawing comparisons between government and corporations, they're probably doing to ignore at least one major difference in order to paint government in a bad light. Not always, but usually.
  - maxerickson 7y ago
    In most corporations, the CEO takes direction from and serves at the leisure of the board (and if there isn't a board, there will typically be some other mechanism where the owners exert control). They typically have a broad mandate (which I suppose is what you are getting at), but they don't literally have unfettered control of the company. (CEOs that own majority stakes in their companies get close to complete control, but that's not what you are arguing, that control flows from the ownership, not the executive position).

pjc50 7y ago

Politics isn't an engineering process, it's a knife fight carried out with pens.

bordercases 7y ago
"I saw him kill three men with a pencil. A fucking. Pencil."

dragonwriter 7y ago

> Imagining legislative text as if it were software code As a software developer with a political science degree and some legal education, you basically shouldn't do this. > this is akin to discovering that an engineer Legislators and executives are even less like engineers in intended function than legal text is like software code. Legislative text isn't software code.b ,

CameronNemo 7y ago
I'm also a swe with a polisci degree (but no legal education or qualifications). I think the comparison between policy and code is useful, but not perfect. Are there any specific areas that are different between them that we should pay mind to?
- dragonwriter 7y ago
  There's a lot, but a couple key points: (1) Legal text is often (arguably, most often in the US system) intended to be reified in application and/or in subordinate enactments (e.g., regulation); not only is it not executable code, it's often not intended as a complete specification. (2) executive and legislative authorities (at least, in the US system) are not like members of a single team, but by-design adversarial power centers. (They are more like customers with conflicting demands than engineers on a team.)