I like this idea a lot. Also I happen to be a member of a community of practice that uses a large group dialogue system that’s in the same category as SDD and Syntegrity. Some colleagues have been playing with using LLMs to enable the work (eg synthesizing summaries from group conversations). I’m going to share this piece with the community to see if I can spark some interest.
I have a similar idea of an online direct democracy, and I found that implementing it is the hardest part. I found a simple way we can do this and if you are willing, we can have a discussion inbox, I just texted you.
Well, the specific methodology for that is Structured Dialogic Design, which I mention in the piece. It was stress tested during negotiations between the Greek and Turkish Cypriots. Like I say, there are actually well-designed and tested approaches to working with deeply divided communities; they’re just not well known.
What I like most about this idea is that it’s actionable. We can start tomorrow. It has the potential to make a wealth of largely latent social technology wisdom available for broad use. I would LOVE to see it actually happen.
Learn how to run and train open-source LLMs. I'm doing that right now--today I'm looking into AnythingLLM. Second, compile a list of documents that make up the training set. Which ones are unavailable in digital versions?
Why not? I know a lot of technically minded people read this newsletter. But so do designers and policy people. It needs to be a collaborative effort because it's inherently interdisciplinary. And I know I have neither the time nor the expertise to do it myself.
I’m fully supportive. Just confirming my understanding and starting to think ahead to practicalities. I honestly don’t think it could or should be done any way but collaboratively.
Yes I would but I am wondering how I might do this practically. Where is the opportunity for this in my local community or one of my virtual communities. An open question for me. Actually I know a community garden full of volunteers with a great space considering using Sociocracy which would potentially be interesting. It is also owned by the NHS so that adds a spicy sauce to the dish. Key consideration - they have limited time/capacity so how would it solve for that...
It was both in the sense that it was an alternative to Soviet-style central planning, which was a system of political as well as economic control. The intent of viable systems is to devolve power to the people closest to its effects.
Back in the 1980s, Stafford Beer gave a lecture to the MBA course I was attending on his cybernetic system and the CuberSyn system he was building for Allende. While the Russian system was "Command and Control", and hopelessly unable to manage complexity, the nested model Beer proposed solved the complexity problem (Ashby law of Requisite Variety", but it was still a hierarchical command system, just as any corporation uses. He made a point that when he asked Allende, who was this system with him at the head nested into, Allende said: "The people" - ie implying democracy.
As an economic model, I don't see it as superior to our distributed, non-hierarchical market system. However, it does offer a model for our political system. In the UK, for example, political control has become more centralized at Westminster, a trend that the current Labour government wants to reverse. Devolution of the countries that make up the UK has also been done to some extent, albeit with ultimate control by Parliament. The US has a federal system where States have considerable power to make decisions but within the bounds of federal law. I only lived in Canada for 1 year back in the late 1970s so I don't know the details of how Canada's provinces and government in Ottawa relate to the US system. What I think is clear that the same tensions in centralization vs independently managed divisions in corporations also appear in political systems. The pendulum swings back and forth on who has control. We even saw this in computing architecture, with the cloud platforms now dominating after various arguments over how peripheral computers would intercommunicate.
While I like the idea of a cybernetic system that could achieve a better democracy, I don't believe it is in any way a silver bullet. The delved Iowa caucus mechanism seems more democratic, but it does seem to fail badly on occasion because there are so many influences pushing people to decide in a particular way. A system that is truly deliberative at the local level would be nice, but unless it truly takes into account the wishes of the population after the issues are properly examined and debated, then it will be subject to the same problems our current very imperfect political system works. A good test would be if a system would ensure that governments reversed course on climate change and pushed hard to try to reverse, or even mitigate the trends (without nuclear bombs being used ;-) as a solution. ) Solutions that would be sustained by all countries, especially the large nations with the most GHG emissions.
Determining who all the affected stakeholders are has always seemed like the hard problem to me. In Stealing Worlds, I used larping as a social form of 'strangemaking' or reframing mechanism, which forced people to look past the assumed categories of their situation. Without that, you face a situation where some hegemony gets to define what (and who) is real. It doesn't matter how efficient your decision-making system is if it ossifies groupthink or ideology in some way. So, for both representative and cybernetic management, I think there is a missing layer of problem definition. In essence, how can we design around David Graeber's idea that our social systems are not monolithic but consist of many overlapping realities, most suppressed at any given moment by some kind of hegemony?
"social systems are not monolithic but consist of many overlapping realities, most suppressed at any given moment by some kind of hegemony?"
Democracy is about suppressing all but the most common POV, restrained by laws. Beer's nested control structure enforces that. Worse, like nested shell companies, minority ownership can even subvert the majority of shareholders in lower layers. To give a concrete example of where political control fails (and democracy wins by subverting the goals set from the leadership) is the UK's goals on housebuilding. Local NIBY councils can block housebuilding to suit residents most in Tory shires. The same shire counties managed to block wind turbines. Is this bad as it subverts political and economic goals, or good that local democracy works? Local stakeholders might gain at the expense of a wider level of stakeholders. A counter-example is the problem of "Cancer Alley" in Louisiana where locals are subjected to serious pollution in favor of the wider state-level position on economic growth.
The best example I recall is the 1990 case where Senatorf Paul Wellstone invited members of the public to listen to various healthcare proposals from different proponents to determine which might be the best to fix the US healthcare system. After days of careful presentations, the panel overwhelmingly chose the Canadian System. Yet as we know, industry scuppered Hillary Clinton's related proposal in the first Clinton administration, and any form of H/C system that is publically financed, even Medicare for US seniors is constantly under attack by conservatives. If the country could have been given the same information, and a referendum passed for this system, would it have been able to overcome a resistant Congress?
IDK what the poor voter turnout in most Democracies is caused by, although I think the lack of difference between political parties is a problem, and the sense that "nothing changes" whoever is voted for. Democracy must have consequences. I find it disheartening as an engaged voter that governments find ways to block public choices. We use representative democracy, but I don't believe our representatives ever read or be advised by staff or experts on the many bills they are asked to vote for. As studies have shown, donors get what they want in preference to the voters.
I would dearly like to read about how democracy could be better used, perhaps by having voters better informed by means as you suggest. However, I suspect that our societies are now too complex to be managed this way, except in momentous cases, as the complexity is overwhelming. Could AI be helpful here, cutting through fabrications and biases and exposing the consequences of a proposal or bill? AI could cut through the legalese of a bill, explain the changes to existing law, and explain the consequences to the reader depending on their situation. This might help overcome the "low-information" voters' tribal position. It will not alleviate coercive action against their representatives, whether from donors, party, or leadership.
IDK if this would be a "wicked problem" as you have talked about periodically, but it is [very] difficult to solve, and may not be solvable for larger population polities. It certainly is failing to solve various global problems. I would welcome experiments to determine what might work best. My personal, naive thought, is that we should ban "professional" representatives and have revolving participation by members of the public informed about specific issues to deliberate. This might both reduce biased influence as well as ensure focus on discrete problems that reduce cognitive overload when all the decisions must pass through too few deciders - an approach that acknowledges human cognitive capabilities (albeit enhanced with AI).
Now we're getting perilously close to the next-gen political worldbuilding that I'm doing for my new novel. There is so much to say here... and so much I should reserve for that book. I have ideas, but they need to cohere--and that's a little hard now as the goalposts for what's considered a plausible future keep being moved.
The challenge with having something in an AI that isn't available as source text is that you can't train the AI to provide you with a deep link into the source text that will let you read the thing it's talking about to make sure it isn't hallucinating.
who has time to read, these days? Well, we should make time. Remember those health and fitness ads that used to so correctly say you can’t afford not to make time/you’re a fool if you don’t etc. To me the same is true of taking time to learn and to ponder. Reading is time travel Carl Sagan said. Our only real chance to live outside our own world. It’s more needed than ever. Now I’m thinking of that old TV show “patience young grasshopper “. 😂😂
Yeah… it would be great if people just did what they should. In the absence of that, we have institutions. These don’t have to be brick-and-mortar, they can be informal and ad-hoc, appearing and disappearing as needed—but they function as “eyeglasses” to correct for the sigmatism of our social skills.
Those institutions can change. My parents used to drink and then drive without seat belts, and say it was okay, then suddenly they didn’t anymore. My Dad’s pipe lay totally unused overnight after he developed a lump on his tongue and the doctor scolded him.
Those moments are interesting, how do we make the institutions as you call them adjust our stigmas?
What we really need the AI to do is to provide citizens and policymakers with tools for seeing political processes better. LLMs can crunch through a lot of data and plot connections between policymakers' actions and outcomes on the local level. An AI seeing tool could also show all of the financial connections that lead to those policymakers.
Both of these would be powerful tools for democratic reform. Policymakers would be able to design policies that had more predictable outcomes and to explain those policies to their constituents. Constituents would be able have a clear idea of who was influencing their policymaker and determine whether or not to vote for them going forward. Seeing the connections between interests will help them align.
Transparency is the key to the future. Making sense of information overload in a complex society is a primary facet of our current malaise. I blogged about an aspect of this issue recently: https://ideaspaces.net/escapingthesilo/
Not incompatible goals, even for one LLM. Just remember GIGO: Garbage In, Garbage Out. The AI can only correctly analyze the political situation if it can get at accurate data. Before using it in that capacity, I would want to rebuild a reliable journalistic class.
Thanks for the quick reply. I agree that journalists are a key pieces of any information ecosystem solution.
There are systemic pressures that make building a "reliable journalistic class" very difficult in today's information environment. I think there are plenty of good journalists out there doing heroic work but they are overwhelmed by the combination of deadline/production pressures and information overload. It's easy to hide stuff in a cemetery if you keep everyone going through the wrong graves.
Journalists need better tools to decipher what's missing from the narrative and use that to triangulate on information (and misinformation). A visual tool for mapping information would do a lot to enable them to do their jobs better and faster. The same tool I described above would work for that purpose as well.
I like this idea a lot. Also I happen to be a member of a community of practice that uses a large group dialogue system that’s in the same category as SDD and Syntegrity. Some colleagues have been playing with using LLMs to enable the work (eg synthesizing summaries from group conversations). I’m going to share this piece with the community to see if I can spark some interest.
This is a fascinating and timely proposal. Decentralized, accessible decision-making tools could be transformative. Great read!
I have a similar idea of an online direct democracy, and I found that implementing it is the hardest part. I found a simple way we can do this and if you are willing, we can have a discussion inbox, I just texted you.
I like this idea, and am especially keen to see how it handles deeply divided polities.
Well, the specific methodology for that is Structured Dialogic Design, which I mention in the piece. It was stress tested during negotiations between the Greek and Turkish Cypriots. Like I say, there are actually well-designed and tested approaches to working with deeply divided communities; they’re just not well known.
What I like most about this idea is that it’s actionable. We can start tomorrow. It has the potential to make a wealth of largely latent social technology wisdom available for broad use. I would LOVE to see it actually happen.
So how can we get started?
Learn how to run and train open-source LLMs. I'm doing that right now--today I'm looking into AnythingLLM. Second, compile a list of documents that make up the training set. Which ones are unavailable in digital versions?
When I read your post, it seemed like you were proposing a collaborative effort. Are you envisioning a collaboratively assembled training set?
Why not? I know a lot of technically minded people read this newsletter. But so do designers and policy people. It needs to be a collaborative effort because it's inherently interdisciplinary. And I know I have neither the time nor the expertise to do it myself.
I’m fully supportive. Just confirming my understanding and starting to think ahead to practicalities. I honestly don’t think it could or should be done any way but collaboratively.
Yes I would but I am wondering how I might do this practically. Where is the opportunity for this in my local community or one of my virtual communities. An open question for me. Actually I know a community garden full of volunteers with a great space considering using Sociocracy which would potentially be interesting. It is also owned by the NHS so that adds a spicy sauce to the dish. Key consideration - they have limited time/capacity so how would it solve for that...
I believe Beer's Chile experience was about managing the economy, not Democracy.
It was both in the sense that it was an alternative to Soviet-style central planning, which was a system of political as well as economic control. The intent of viable systems is to devolve power to the people closest to its effects.
Back in the 1980s, Stafford Beer gave a lecture to the MBA course I was attending on his cybernetic system and the CuberSyn system he was building for Allende. While the Russian system was "Command and Control", and hopelessly unable to manage complexity, the nested model Beer proposed solved the complexity problem (Ashby law of Requisite Variety", but it was still a hierarchical command system, just as any corporation uses. He made a point that when he asked Allende, who was this system with him at the head nested into, Allende said: "The people" - ie implying democracy.
As an economic model, I don't see it as superior to our distributed, non-hierarchical market system. However, it does offer a model for our political system. In the UK, for example, political control has become more centralized at Westminster, a trend that the current Labour government wants to reverse. Devolution of the countries that make up the UK has also been done to some extent, albeit with ultimate control by Parliament. The US has a federal system where States have considerable power to make decisions but within the bounds of federal law. I only lived in Canada for 1 year back in the late 1970s so I don't know the details of how Canada's provinces and government in Ottawa relate to the US system. What I think is clear that the same tensions in centralization vs independently managed divisions in corporations also appear in political systems. The pendulum swings back and forth on who has control. We even saw this in computing architecture, with the cloud platforms now dominating after various arguments over how peripheral computers would intercommunicate.
While I like the idea of a cybernetic system that could achieve a better democracy, I don't believe it is in any way a silver bullet. The delved Iowa caucus mechanism seems more democratic, but it does seem to fail badly on occasion because there are so many influences pushing people to decide in a particular way. A system that is truly deliberative at the local level would be nice, but unless it truly takes into account the wishes of the population after the issues are properly examined and debated, then it will be subject to the same problems our current very imperfect political system works. A good test would be if a system would ensure that governments reversed course on climate change and pushed hard to try to reverse, or even mitigate the trends (without nuclear bombs being used ;-) as a solution. ) Solutions that would be sustained by all countries, especially the large nations with the most GHG emissions.
Determining who all the affected stakeholders are has always seemed like the hard problem to me. In Stealing Worlds, I used larping as a social form of 'strangemaking' or reframing mechanism, which forced people to look past the assumed categories of their situation. Without that, you face a situation where some hegemony gets to define what (and who) is real. It doesn't matter how efficient your decision-making system is if it ossifies groupthink or ideology in some way. So, for both representative and cybernetic management, I think there is a missing layer of problem definition. In essence, how can we design around David Graeber's idea that our social systems are not monolithic but consist of many overlapping realities, most suppressed at any given moment by some kind of hegemony?
"social systems are not monolithic but consist of many overlapping realities, most suppressed at any given moment by some kind of hegemony?"
Democracy is about suppressing all but the most common POV, restrained by laws. Beer's nested control structure enforces that. Worse, like nested shell companies, minority ownership can even subvert the majority of shareholders in lower layers. To give a concrete example of where political control fails (and democracy wins by subverting the goals set from the leadership) is the UK's goals on housebuilding. Local NIBY councils can block housebuilding to suit residents most in Tory shires. The same shire counties managed to block wind turbines. Is this bad as it subverts political and economic goals, or good that local democracy works? Local stakeholders might gain at the expense of a wider level of stakeholders. A counter-example is the problem of "Cancer Alley" in Louisiana where locals are subjected to serious pollution in favor of the wider state-level position on economic growth.
The best example I recall is the 1990 case where Senatorf Paul Wellstone invited members of the public to listen to various healthcare proposals from different proponents to determine which might be the best to fix the US healthcare system. After days of careful presentations, the panel overwhelmingly chose the Canadian System. Yet as we know, industry scuppered Hillary Clinton's related proposal in the first Clinton administration, and any form of H/C system that is publically financed, even Medicare for US seniors is constantly under attack by conservatives. If the country could have been given the same information, and a referendum passed for this system, would it have been able to overcome a resistant Congress?
IDK what the poor voter turnout in most Democracies is caused by, although I think the lack of difference between political parties is a problem, and the sense that "nothing changes" whoever is voted for. Democracy must have consequences. I find it disheartening as an engaged voter that governments find ways to block public choices. We use representative democracy, but I don't believe our representatives ever read or be advised by staff or experts on the many bills they are asked to vote for. As studies have shown, donors get what they want in preference to the voters.
I would dearly like to read about how democracy could be better used, perhaps by having voters better informed by means as you suggest. However, I suspect that our societies are now too complex to be managed this way, except in momentous cases, as the complexity is overwhelming. Could AI be helpful here, cutting through fabrications and biases and exposing the consequences of a proposal or bill? AI could cut through the legalese of a bill, explain the changes to existing law, and explain the consequences to the reader depending on their situation. This might help overcome the "low-information" voters' tribal position. It will not alleviate coercive action against their representatives, whether from donors, party, or leadership.
IDK if this would be a "wicked problem" as you have talked about periodically, but it is [very] difficult to solve, and may not be solvable for larger population polities. It certainly is failing to solve various global problems. I would welcome experiments to determine what might work best. My personal, naive thought, is that we should ban "professional" representatives and have revolving participation by members of the public informed about specific issues to deliberate. This might both reduce biased influence as well as ensure focus on discrete problems that reduce cognitive overload when all the decisions must pass through too few deciders - an approach that acknowledges human cognitive capabilities (albeit enhanced with AI).
Now we're getting perilously close to the next-gen political worldbuilding that I'm doing for my new novel. There is so much to say here... and so much I should reserve for that book. I have ideas, but they need to cohere--and that's a little hard now as the goalposts for what's considered a plausible future keep being moved.
Then I look forward to reading your next novel and what you have to say about this.
Is this the Platform for Change you were looking for? https://www.amazon.com/Talking-Point-Thomas-R-Flanagan-ebook/dp/B09P42T422/
The challenge with having something in an AI that isn't available as source text is that you can't train the AI to provide you with a deep link into the source text that will let you read the thing it's talking about to make sure it isn't hallucinating.
Good point.
who has time to read, these days? Well, we should make time. Remember those health and fitness ads that used to so correctly say you can’t afford not to make time/you’re a fool if you don’t etc. To me the same is true of taking time to learn and to ponder. Reading is time travel Carl Sagan said. Our only real chance to live outside our own world. It’s more needed than ever. Now I’m thinking of that old TV show “patience young grasshopper “. 😂😂
Yeah… it would be great if people just did what they should. In the absence of that, we have institutions. These don’t have to be brick-and-mortar, they can be informal and ad-hoc, appearing and disappearing as needed—but they function as “eyeglasses” to correct for the sigmatism of our social skills.
Those institutions can change. My parents used to drink and then drive without seat belts, and say it was okay, then suddenly they didn’t anymore. My Dad’s pipe lay totally unused overnight after he developed a lump on his tongue and the doctor scolded him.
Those moments are interesting, how do we make the institutions as you call them adjust our stigmas?
What we really need the AI to do is to provide citizens and policymakers with tools for seeing political processes better. LLMs can crunch through a lot of data and plot connections between policymakers' actions and outcomes on the local level. An AI seeing tool could also show all of the financial connections that lead to those policymakers.
Both of these would be powerful tools for democratic reform. Policymakers would be able to design policies that had more predictable outcomes and to explain those policies to their constituents. Constituents would be able have a clear idea of who was influencing their policymaker and determine whether or not to vote for them going forward. Seeing the connections between interests will help them align.
Transparency is the key to the future. Making sense of information overload in a complex society is a primary facet of our current malaise. I blogged about an aspect of this issue recently: https://ideaspaces.net/escapingthesilo/
Not incompatible goals, even for one LLM. Just remember GIGO: Garbage In, Garbage Out. The AI can only correctly analyze the political situation if it can get at accurate data. Before using it in that capacity, I would want to rebuild a reliable journalistic class.
Thanks for the quick reply. I agree that journalists are a key pieces of any information ecosystem solution.
There are systemic pressures that make building a "reliable journalistic class" very difficult in today's information environment. I think there are plenty of good journalists out there doing heroic work but they are overwhelmed by the combination of deadline/production pressures and information overload. It's easy to hide stuff in a cemetery if you keep everyone going through the wrong graves.
Journalists need better tools to decipher what's missing from the narrative and use that to triangulate on information (and misinformation). A visual tool for mapping information would do a lot to enable them to do their jobs better and faster. The same tool I described above would work for that purpose as well.