4 ms·

The heart of the problem is that popular communication networks are centralized and unencrypted.

by thomastjeffery 9y ago

tlb 9y ago
How would you make a decentralized, encrypted twitter? How would it be better for society?
- dredmorbius 9y ago
  https://github.com/tootsuite/mastodon https://github.com/tootsuite/mastodon
- thomastjeffery 9y ago
  It would not be Twitter. Twitter is inherently public. The issue is that people flock to public platforms to share personal data. What we need is the equivalent of Email and GPG, but actually user-friendly.
dredmorbius 9y ago
Good protocols don't cure bad ethics. Mind: I would like to see decentralisation and better encryption protections, generally, but those by themselves won't solve the problem (you still need to address ethics, rights, and power imbalances), and still leave certain of present attacks in place. David Gerard, author of Attack of the 50 Foot Blockchain, makes the exceedingly good point in an FT interview, that Bitcoin and Blockchain themselves are attempts to tech around the trust problem, and that this raises challenges because trust itself buys you an immense amount of efficiency. https://www.ft.com/content/61cdc5c8-370e-11e7-bce4-9023f8c0fd2e https://www.ft.com/content/61cdc5c8-370e-11e7-bce4-9023f8c0f... I'm starting to come to the view that information technology itself directly attacks* interpersonal trust, at the macro scale. I may not be articulating the argument well at this point, though I've tried: https://www.reddit.com/r/dredmorbius/comments/6jqakv/communications_advances_undermine_trust/ https://www.reddit.com/r/dredmorbius/comments/6jqakv/communi...
- thomastjeffery 9y ago
  > cure bad ethics. I don't see "bad ethics" as a disease, but a difference of opinion. Why should I compel others to share my opinion?
- TeMPOraL 9y ago
  Not sure if the technology itself attacks interpersonal trust - but I do see a lot work in tech being done exactly towards that. In particular, crypto folks from the blockchain sphere do their damnest to invent ways in which you don't have to trust anybody to do things. I do not feel it's a good idea, at least not for basing society and its infrastructure on that. The core of my argument would be: trust is something we're good at, as humans. It's efficient, and it's flexible. By flexibility I mean that it handles new/weird situations well. Trustless solutions, on the other hand, burn a lot of compute to replace the need to trust another person, and they do not handle corner cases in any way. A similar thing is bureaucracy - it exists as a way to reduce dependence on trust, at a serious loss of inefficiency. Frankly, I'd argue that today's bureaucracies are so inefficient that without hidden trust acting as a grease on the wheels, they'd grind to a complete halt. They're also inflexible, and they're usually only saved - again - through trust that leaves room for individual bureaucrats to "cheat the system" here and there. All in all, I feel the problem is with scale of societies we're trying to create. When it was just 100 people in your village tribe, everyone knew everyone else, and trust-based society just worked. As we grew our societies by many orders of magnitude, we've lost the interpersonal mechanisms of establishing and maintaining trust (which includes punishing actors for breaking it). And it seems to me that instead of trying to find new mechanisms for enabling trust at scale, we're just giving up on the whole idea and replacing it with burning physical resources. This, I feel, is a wrong way to go. (I'm probably not articulating the argument well either.)
  - dredmorbius 9y ago
    Take a look at the Reddit item I linked, it's my own (uncharacteristically brief, you're welcome) essay. I've been digging into literate on this further, and ... there's some general support, though the overall picture's somewhat mixed. My starting observation was that, while the ability for two parties to voluntarily communicate might increase their trust, the ability for a party (or parties) to unilaterally surveil others, particularly at scale, is almost certainly corrosive to trust. Issues such as comprehensive systems surveillance (user logging and monitoring on computers), cameras, microphones, etc., come to mind. In reading on the topic, there seems to be a great deal in sociology, particularly Durkheim and Weber, addressing the point. I'd observed that every major empire has had a strong religious component (with the possible exception of the Mongol Horde), and that religion didn't start breaking down strongly until ~18th and 19th centuries, under the onslaught of both science and reason, and improved communications. JoAnne Yates and James Beniger have both written extensively on the nature of communications technologies and practices especially within commercial contexts. Beniger's The Control Revolution in particular details the evolution of commercial and commodities practices in the Americas starting before the American Revolution. A significant feature especially prior to the 1830s / 1840s, at which point the telegraph provided instant communications especially of prices was that movements of goods required agents, working at a distance, and with considerable independence and autonomy. Beniger describes the counterflows of goods and financial records, generated at each transshippment point (each triggering a transfer of goods, a financial transaction, and a trust relationship between originator, buyer, and agents). I'm not saying that the trust was always well-placed, but that there was no real alternative, other than setting up rules, perhaps some form of non-realtime checks (multiple agents rather than one, say, such that outlier behaviour might be observed), and a very strong reliance on reputation. This is reflected even in the language of business communications, which is based on establishing and maintaining trust -- later streamlining of external and internal communications in the early 20th century removed much of the ornate 19th century language and replaced it with the no-nonsense, strictly-business correspondence of the 20th century. Email, texts, and Tweets have stripped that further still. Beniger specifically mentions bureaucracy as a technology, by the way. And yes, scale has a ton to do with this. Another notion I'm looking at is that networks (I'm looking at interpersonal / social, as well as technological) directly affect information flows through size, topology, cost functions, cohort selection, protocols, and more. (Yates appears to be studying a fair bit of this, though I'm not sure to what extent her views are as technical or information-theoretic as mine.) Bureaucracies are, by the way, generally very efficient, so long as they can be subject to efficiency constraints. They do get hidebound, in that a bureaucracy is a mechanism for formalising information flows (literally "creating forms" -- a major class of business correspondence), and that there's always a problem between a tightly-specified limited grammar and the Real World. (See: things programmers believe about time, time zones, names, places, etc., etc., etc.) There's also the matter of dealing with information at scale, for which I find powers of ten a useful-if-rough organising concept. Treat these as ranges, with [previous bound] ~< n ~<= [stated upper bound], and yes, with fuzzy edges. 10^0: Full focus. An item. 10^1: Reasonable scope of daily focus. A short list. 10^2: An information-gathering scope, it's possible to be at least generally aware of each item. A long list. 10^3: Stressing the bounds of even skilled individual un-managed awareness. A compilation of lists, a book. Pre-Gutenberg Europe had about 30,000 total books (volumes, not titles), and few libraries exceeded 1,000 total volumes. 10^4 - 10^5: Some sort of organised management system is necessary, can be paper-based. About 300k - 1 million books are published annually ("traditional" and "self-published" respectively). 10^6 - 10^8: A formalised and electronic system is probably necessary at this point. Statistical treatments quite helpful. The largest libraries contain ~24 million volumes. 10^9 - 10^11: Statistical treatments almost certainly necessary. 10^12+: Big data, machine-learning. Multiple levels of abstraction required to humanly comprehend scales. A galaxy's worth of stars. Total number of cells in the human body. I use books, notes, and index cards considerably. The sense of information scale (and tractability) as I range from 1 card to 10 to 100 to 1,000s is interesting and visceral. Now make those people and organisations and businesses and systems, and try to work with them .... See: Beniger: http://www.worldcat.org/title/control-revolution-technological-and-economic-origins-of-the-information-society/oclc/895626594 http://www.worldcat.org/title/control-revolution-technologic... Yates: http://www.worldcat.org/title/information-technology-and-organizational-transformation-history-rhetoric-and-practice/oclc/44669262 http://www.worldcat.org/title/information-technology-and-org...

The heart of the problem is that popular communication networks are centralized and unencrypted.

tlb 9y ago

How would you make a decentralized, encrypted twitter? How would it be better for society?

dredmorbius 9y ago
https://github.com/tootsuite/mastodon https://github.com/tootsuite/mastodon
thomastjeffery 9y ago
It would not be Twitter. Twitter is inherently public. The issue is that people flock to public platforms to share personal data. What we need is the equivalent of Email and GPG, but actually user-friendly.

dredmorbius 9y ago

Good protocols don't cure bad ethics. Mind: I would like to see decentralisation and better encryption protections, generally, but those by themselves won't solve the problem (you still need to address ethics, rights, and power imbalances), and still leave certain of present attacks in place. David Gerard, author of Attack of the 50 Foot Blockchain, makes the exceedingly good point in an FT interview, that Bitcoin and Blockchain themselves are attempts to tech around the trust problem, and that this raises challenges because trust itself buys you an immense amount of efficiency. https://www.ft.com/content/61cdc5c8-370e-11e7-bce4-9023f8c0fd2e https://www.ft.com/content/61cdc5c8-370e-11e7-bce4-9023f8c0f... I'm starting to come to the view that information technology itself directly attacks* interpersonal trust, at the macro scale. I may not be articulating the argument well at this point, though I've tried: https://www.reddit.com/r/dredmorbius/comments/6jqakv/communications_advances_undermine_trust/ https://www.reddit.com/r/dredmorbius/comments/6jqakv/communi...

thomastjeffery 9y ago
> cure bad ethics. I don't see "bad ethics" as a disease, but a difference of opinion. Why should I compel others to share my opinion?
TeMPOraL 9y ago
Not sure if the technology itself attacks interpersonal trust - but I do see a lot work in tech being done exactly towards that. In particular, crypto folks from the blockchain sphere do their damnest to invent ways in which you don't have to trust anybody to do things. I do not feel it's a good idea, at least not for basing society and its infrastructure on that. The core of my argument would be: trust is something we're good at, as humans. It's efficient, and it's flexible. By flexibility I mean that it handles new/weird situations well. Trustless solutions, on the other hand, burn a lot of compute to replace the need to trust another person, and they do not handle corner cases in any way. A similar thing is bureaucracy - it exists as a way to reduce dependence on trust, at a serious loss of inefficiency. Frankly, I'd argue that today's bureaucracies are so inefficient that without hidden trust acting as a grease on the wheels, they'd grind to a complete halt. They're also inflexible, and they're usually only saved - again - through trust that leaves room for individual bureaucrats to "cheat the system" here and there. All in all, I feel the problem is with scale of societies we're trying to create. When it was just 100 people in your village tribe, everyone knew everyone else, and trust-based society just worked. As we grew our societies by many orders of magnitude, we've lost the interpersonal mechanisms of establishing and maintaining trust (which includes punishing actors for breaking it). And it seems to me that instead of trying to find new mechanisms for enabling trust at scale, we're just giving up on the whole idea and replacing it with burning physical resources. This, I feel, is a wrong way to go. (I'm probably not articulating the argument well either.)
- dredmorbius 9y ago
  Take a look at the Reddit item I linked, it's my own (uncharacteristically brief, you're welcome) essay. I've been digging into literate on this further, and ... there's some general support, though the overall picture's somewhat mixed. My starting observation was that, while the ability for two parties to voluntarily communicate might increase their trust, the ability for a party (or parties) to unilaterally surveil others, particularly at scale, is almost certainly corrosive to trust. Issues such as comprehensive systems surveillance (user logging and monitoring on computers), cameras, microphones, etc., come to mind. In reading on the topic, there seems to be a great deal in sociology, particularly Durkheim and Weber, addressing the point. I'd observed that every major empire has had a strong religious component (with the possible exception of the Mongol Horde), and that religion didn't start breaking down strongly until ~18th and 19th centuries, under the onslaught of both science and reason, and improved communications. JoAnne Yates and James Beniger have both written extensively on the nature of communications technologies and practices especially within commercial contexts. Beniger's The Control Revolution in particular details the evolution of commercial and commodities practices in the Americas starting before the American Revolution. A significant feature especially prior to the 1830s / 1840s, at which point the telegraph provided instant communications especially of prices was that movements of goods required agents, working at a distance, and with considerable independence and autonomy. Beniger describes the counterflows of goods and financial records, generated at each transshippment point (each triggering a transfer of goods, a financial transaction, and a trust relationship between originator, buyer, and agents). I'm not saying that the trust was always well-placed, but that there was no real alternative, other than setting up rules, perhaps some form of non-realtime checks (multiple agents rather than one, say, such that outlier behaviour might be observed), and a very strong reliance on reputation. This is reflected even in the language of business communications, which is based on establishing and maintaining trust -- later streamlining of external and internal communications in the early 20th century removed much of the ornate 19th century language and replaced it with the no-nonsense, strictly-business correspondence of the 20th century. Email, texts, and Tweets have stripped that further still. Beniger specifically mentions bureaucracy as a technology, by the way. And yes, scale has a ton to do with this. Another notion I'm looking at is that networks (I'm looking at interpersonal / social, as well as technological) directly affect information flows through size, topology, cost functions, cohort selection, protocols, and more. (Yates appears to be studying a fair bit of this, though I'm not sure to what extent her views are as technical or information-theoretic as mine.) Bureaucracies are, by the way, generally very efficient, so long as they can be subject to efficiency constraints. They do get hidebound, in that a bureaucracy is a mechanism for formalising information flows (literally "creating forms" -- a major class of business correspondence), and that there's always a problem between a tightly-specified limited grammar and the Real World. (See: things programmers believe about time, time zones, names, places, etc., etc., etc.) There's also the matter of dealing with information at scale, for which I find powers of ten a useful-if-rough organising concept. Treat these as ranges, with [previous bound] ~< n ~<= [stated upper bound], and yes, with fuzzy edges. 10^0: Full focus. An item. 10^1: Reasonable scope of daily focus. A short list. 10^2: An information-gathering scope, it's possible to be at least generally aware of each item. A long list. 10^3: Stressing the bounds of even skilled individual un-managed awareness. A compilation of lists, a book. Pre-Gutenberg Europe had about 30,000 total books (volumes, not titles), and few libraries exceeded 1,000 total volumes. 10^4 - 10^5: Some sort of organised management system is necessary, can be paper-based. About 300k - 1 million books are published annually ("traditional" and "self-published" respectively). 10^6 - 10^8: A formalised and electronic system is probably necessary at this point. Statistical treatments quite helpful. The largest libraries contain ~24 million volumes. 10^9 - 10^11: Statistical treatments almost certainly necessary. 10^12+: Big data, machine-learning. Multiple levels of abstraction required to humanly comprehend scales. A galaxy's worth of stars. Total number of cells in the human body. I use books, notes, and index cards considerably. The sense of information scale (and tractability) as I range from 1 card to 10 to 100 to 1,000s is interesting and visceral. Now make those people and organisations and businesses and systems, and try to work with them .... See: Beniger: http://www.worldcat.org/title/control-revolution-technological-and-economic-origins-of-the-information-society/oclc/895626594 http://www.worldcat.org/title/control-revolution-technologic... Yates: http://www.worldcat.org/title/information-technology-and-organizational-transformation-history-rhetoric-and-practice/oclc/44669262 http://www.worldcat.org/title/information-technology-and-org...