[HN Gopher] Centralization, Decentralization, and Internet Stand...
___________________________________________________________________
Centralization, Decentralization, and Internet Standards
Author : mooreds
Score : 62 points
Date : 2023-12-30 11:20 UTC (11 hours ago)
(HTM) web link (www.rfc-editor.org)
(TXT) w3m dump (www.rfc-editor.org)
| zaik wrote:
| Past discussion related to this RFC:
| https://news.ycombinator.com/item?id=38694004
| 0xbadcafebee wrote:
| These words ("centralization", "decentralization") have become a
| cargo-cult phenomenon. People are imbuing them with meaning they
| don't have, and trying to use them to solve a problem I don't
| think they really understand.
|
| _" In this document, "centralization" is the state of affairs
| where a single entity or a small group of them can observe,
| capture, control, or extract rent from the operation or use of an
| Internet function exclusively."_
|
| There are other terms the authors could use: oligopoly,
| oligarchy, inverted totalitarianism, plutocracy, soft despotism,
| rent-seeking. But they didn't choose these terms, because the
| authors aren't political science majors, they are nerds. They're
| looking at the problem not as one of a larger set of economic,
| political and social factors, but one of a computer science
| problem to be solved with computer science means. If you've only
| ever been a carpenter, and your only tool is a hammer, every
| problem looks like a nail.
|
| But this isn't a computer science problem, and none of the
| solutions presented in this document will prevent the forces at
| work from merely "routing around" any technological barriers
| imposed. You cannot program around a government.
| y8bj wrote:
| You're trying to constrain meaning to your sensibilities. An
| authority _you_ don't have.
|
| Note humanity does not use Latin in day to day life.
|
| Language is not immutable physics.
| dale_glass wrote:
| I disagree, it's in a large part a computer science problem.
|
| So for instance, I work on a decentralized VR world
| (https://overte.org/). Our architecture is very similar to that
| of the web -- we have a 3D analog of a web server, and a 3D
| analog of a web browser. Authentication is like on the web --
| anonymous by default. Storage is wherever you want to fetch
| stuff from by HTTP.
|
| That's technically different from other systems in the same
| space. Eg, Second Life is a very centralized system. There's a
| grid, the various simulators are placed on a map, there's a
| central asset storage, and everyone must log in.
|
| If Second Life open sourced their backend, then the resulting
| outcome would be competing centralized platforms, where you
| either log into SL, or you log into the alternative. Each of
| those is a self-contained universe that you can be completely
| be removed from.
|
| And sure, there are various political implications there, but
| it's also a matter of what technical decisions and tradeoffs
| you make. SL for instance assumes everyone has an user account.
| Our system assumes everyone is anonymous by default. In SL
| there can be a coherent notion of a "world map" because a
| central entity decides which simulator goes where. In our
| system, such a thing is not possible because any given world
| has an arbitrary size and arbitrary owner, there's no way to
| say Alice and Bob are neighbors, nor there is a central system
| to ask for a definitive answer.
|
| The system is so well designed for this that IMO it likely was
| one of the things that made the business model of the original
| maker of it (High Fidelity) fail. The design is just naturally
| resistant to monetization. To the point that when I first
| encountered it, I was baffled at how it was ever going to make
| any money. The very design looks like something a highly
| ideological and anti-corporate group would come up with to make
| sure that even if some of them fell to the "dark side", there
| still wouldn't be a way to make a profit.
| nonrandomstring wrote:
| I am sympathetic to your structural take, and also to the
| parent comment that technologists lack political chops. But I
| lost you here:
|
| > something a highly ideological and anti-corporate group
| would come up with to make sure that even if some of them
| fell to the "dark side", there still wouldn't be a way to
| make a profit.
|
| No. The motive is to make sure that even if some of them fell
| to the "dark side", there wouldn't be a way to destroy the
| efforts of those who choose not to make a profit.
| dale_glass wrote:
| That part is not really serious. The point is that this was
| a commercial company, of a decent size and investment. And
| somehow it produced a design that one would expect to come
| out from some sort of highly anti-corporate group.
|
| Normally a company has a business plan of some kind. This
| design is almost as if somebody put a lot of thought into
| making sure there couldn't be one.
| nonrandomstring wrote:
| I see your point that as an established commercial
| company they apparently failed to diligently research
| their returns.
|
| > somebody put a lot of thought into making sure there
| couldn't be one
|
| Perhaps nobody put thought into that. Maybe they're just
| bad business thinkers. But equating openness to an
| ideological anti-corporate bent seems a little strong. Is
| it possible they saw a profitable path that is also open?
| dale_glass wrote:
| Well, take Second Life, VR Chat, NeosVR, Resonite, etc.
| They all are pretty similar concept-wise. First thing you
| get is a login screen. You need an account. On some, that
| account costs (or used to) cost money.
|
| The world is locked inside a walled garden. If you don't
| like SL's rules, you don't get to touch anything SL
| related. You can't talk to your SL friends without
| logging into SL.
|
| The server code is closed source. You must use the
| central infrastructure. In some cases, like SL, the
| client is open.
|
| They sell add-ons like storage space for your stuff, and
| things like avatars you can directly buy. Some have an
| internal economy, but eg, Second Life quickly moved to
| being the official owners of the internal marketplace,
| and tax transactions.
|
| If you want to have a permanent chunk of the world in SL,
| you pay for that, monthly.
|
| All pretty standard. Now compare High Fidelity.
|
| You install the client, open it, and instantly you show
| up in-world with no login screen required.
|
| For data storage the solution is to pay any random web
| provider, or use the free AWS S3 tier. Any asset is just
| an URL pointing to https://example.com/file.gltf --
| served from any random server, downloaded client-side,
| and outside HiFi's control.
|
| Okay, so you pay for hosting your own environment? Nope.
| You host it on any Windows or Linux host you like, or the
| installer will even automatically install a server
| locally, easily start it from a menu, and NAT hole punch
| so you can host stuff on your laptop if you want to.
|
| Tax Alice selling stuff to Bob? Well, the system is so
| open that anyone can set up their own shop if they want
| to. The source code for the shop is open, so is both the
| server and client. Since assets are hosted externally,
| nothing prevents Alice paying Bob with Paypal, emailing a
| file, and escaping any taxation.
|
| Excluding people who break the rules? No such thing,
| there's no "grid". You can only be banned from individual
| worlds, like HN blocking you off doesn't mean anything
| for Ars Technica.
|
| And all of that was Apache 2.0 licensed, so anybody who
| wanted could absolutely set up their own commercial
| operation and not share any of their own improvements.
|
| If it was me running it, I'd have put the AGPL on it to
| discourage competitors, and either focus completely on
| custom setups for whoever needs one, or try to run a very
| light operation subsisting from Patreon.
| nonrandomstring wrote:
| My take would be that they set out to "minimise
| friction", build a user base and monetise that later, but
| having never seen the "Underpants Gnomes" episode of
| South Park.
|
| One could forgive that because the "build it first,
| squeeze it later" model /has/ worked for other tech
| businesses. Twitter comes to mind.
|
| One could build a "premium" closed ecosystem within an
| open ecosystem which inducts new customers in. I don't
| think the design was fundamentally structurally flawed,
| so much as they didn't follow up on the opportunities
| created by the initial phase.
|
| That said, I think it's a very hard transition, even if
| you did plan well in advance.
| dale_glass wrote:
| > One could forgive that because the "build it first,
| squeeze it later" model /has/ worked for other tech
| businesses. Twitter comes to mind.
|
| Twitter is a walled garden though. This is more like if
| Twitter built Mastodon, and nearly everyone ran their own
| server.
|
| So you what, sell ads? But everyone browses from their
| own install which doesn't serve any.
|
| Part of the squeezing strategy is that you have a captive
| audience that can be suddenly asked to pay, or watch ads
| and they don't want to lose their social network, so they
| do. But this model had no such thing.
|
| What I heard was that their intended model was "tax the
| economy", but besides being trivial to work around in the
| architecture, every content creator I met hated the
| completely wide open architecture. There's no way to
| protect content in the system. Not only it's trivial to
| grab somebody else's stuff, you can't be punished for
| doing so.
| nonrandomstring wrote:
| To answer from a personal point of view, I'm very, very
| old fashioned. I believe that successful long term
| business only comes from providing compelling value to
| customers, building loyalty and getting them to willingly
| pay you for what you offer, not what you threaten to take
| away.
|
| Much of the "capture then exploit" thinking of the recent
| tech world, walled gardens and the like, is anathema to
| me. Once you antagonise your own customers it's a losing
| game. They will circumvent and subvert everything you do,
| and good on them for being smarter and more agile. In
| other words I think that whole mentality is wrong, and
| indeed immoral.
|
| OTOH, that isn't to say you can't build great business in
| other ways, and especially within interoperable, free and
| open systems which also add value to the whole world as
| well as bring you a profitable living.
|
| So for me, "squeezing" is about using the fertile ground
| you create through enabling freedom to grow something
| that's privately profitable. When you don't have to worry
| about "protecting content" and enforcing behaviour you
| can concentrate on creating value.
|
| Such businesses are not super scalable, and won't make
| you a billionaire, or satisfy investors. But they can
| make you a good living and are potentially stable and
| sustainable. I know it doesn't fit the cultural values of
| Silicon Valley, but to me that's the gold standard of
| business.
|
| Also my experience is that it isn't ones own customers,
| or peer competitors that we need protection from. It's
| the Microsoft, Google and Facebooks of the world who can
| steal your lunch with impunity, write laws, break laws...
| they'll steal or destroy whatever they like, so in many
| ways it pays to stay under their radar.
| dwaite wrote:
| I would argue VR is not a neutral example.
|
| In the physical world, space is a limited resource that
| sells at a premium. You have benefit from things being in
| a reasonable distance for walking/traveling. Being near
| water has real-world benefits for transportation and
| trade as well as agriculture.
|
| A virtual world is still based on fundamental spatial
| concepts like needing to travel, having benefit of
| proximity/association, and so on.
|
| So there needs to be a specific set of decisions made on
| how to deal with spatial assumptions when they go counter
| to your design goals:
|
| - Games have instanced areas because otherwise there
| would be physically too many people to move, or allow you
| to walk through other people to get around.
|
| - A chat service might represent _your_ rooms off a hub,
| rather than a megaplex conference center of all rooms.
|
| - Individual zones might be represented as distinct
| 'worlds' with portal entrances, to allow for each to have
| its own spatial rules (and even physics), where each has
| its own sense of layout/locality.
|
| each of these has their own sets of trade-offs, but
| importantly you want consistency for users. That set of
| trade-offs literally lead to representing the world as a
| walled garden.
|
| That is before you get to business interests being first
| and foremost around succeeding with their own products,
| not interoperability with nonexistent or noncooperative
| competitors. The market has to hit a point where vendors
| agree that interoperability is important enough to accept
| the tradeoffs that come with it, before any technical
| effort there can succeed.
| Karellen wrote:
| > There are other terms the authors could use: oligopoly,
| oligarchy, inverted totalitarianism, plutocracy, soft
| despotism, rent-seeking.
|
| Those are motivations behind centralisation, but they won't all
| apply to all attempts at centralisation. There may be some
| situations where _none_ of those particular motivations are the
| apparent reason behind a particular attempt at
| centralisation.[0]
|
| Using any one motivation, rather than the descriptive
| "centralisation", to describe the phenomena will necessarily
| miss instances of centralisation that the discussion should
| apply to. Using a long list of motivations every time would be
| unwieldy, and could still miss instances that should be
| referenced.
|
| [0] e.g. Signal https://signal.org/blog/the-ecosystem-is-
| moving/
| FridgeSeal wrote:
| Agreed. Developers - and by extension communities of them -
| tend to come up with technical solutions to problems,
| regardless of whether the problem is technical in nature, which
| invariably either don't work at all, or make the situation
| _actively worse_ when the technical solution is weaponised by
| those at fault in the first place.
|
| I'm not sure how we address this, which is a pity because even
| an uncertain response would be better than the "confident
| technical solution" we often get thrown around.
| jmyeet wrote:
| I tend to have a sinking feeling whenever this topic comes up
| because there's a lot of religious fervor around federation in
| particular that is completely divorced from reality and
| practicality.
|
| But, the section of this on federation is remarkably level-headed
| and pragmatic. It's not a panacea and it creates a lot of
| friction and new problems that otherwise wouldn't exist.
|
| And this is before you get into the benefits for the end-user,
| which are typically none.
|
| The profit motive will always build a wall--technically,
| operationally--or an "enclosure", if you will, legislatively or
| politically. The only ones committed to "openness" or "open
| source" or "open access" are those who have already lost in the
| marketplace, which is why people say things like "open source is
| for losers" [1].
|
| If we look at what works it's collective ownership eg Wikipedia.
| As another commenter shrewdly noted, this is a political not a
| technical problem. It's about the workers relationship to the
| means of production. It's who gets to capture ("exploit") the
| value created. And that's not a political statement. It's just
| materialist analysis.
|
| But people don't want to hear that because they see themselves as
| future bag-holders and collective ownership threatens their
| ability to one day be at the top of the ladder, even though (most
| likely) they'll never be at the top of the ladder.
|
| [1]: https://siliconangle.com/2014/05/29/only-loser-vendors-
| are-t...
___________________________________________________________________
(page generated 2023-12-30 23:00 UTC)