sofiechan home

The third-worldification of the global internet: are private network enclaves the answer?

anon_voge said in #3574 7d ago: received

Following recent anon discourse on the general third-worldification of the internet, I've been pondering potential solutions.

Is there a valid case for the emergence of private, invite-only network enclaves? Projects like Yggdrasil [1] give a solid example for an overlay network made up of participant nodes operating on top of existing physical networks ("the internet"). It would be perfectly feasible to add some extra gating mechanisms on top--things such as requiring invites or financial contribution--to build a 'members only' internet.

[1] https://yggdrasil-network.github.io/

referenced by: >>3605

Following recent ano received

phaedrus said in #3577 7d ago: received

Generally, sorting out high signal contributions from all of the insane third-world trash is a really worthwhile endeavor. However, more private networks with higher barriers to entry suffer from the lack of network effects and a reduced openness to talented contributors that can be fatal. I think the structural barriers to posting on SofieChan are pretty good, and it seems to be spreading among the desired intellectual set. However, if it becomes more tightly locked down, that will have a negative influence on the incoming trickle of high-value posters. So something like cultural and structural barriers, like an imposing interface and longer-form writing, seem like a sweet spot that excludes the right people while remaining open to interesting new talent.

referenced by: >>3587

Generally, sorting o received

anon_betu said in #3581 7d ago: received

As the internet becomes overwhelmed with diversity, the masses, and AI slop, it does become necessary to construct more selective subnets of various kinds. At a most basic level, social networks which do this or provide for it will become more valuable. This is arguably what reddit and twitter are about. Now of course we don't totally like their particular curation or lack of curation, and maybe that points to a market opportunity. But to first approximation, this is a self-solving problem because of the value opportunity in solving it.

I think people reach too quickly to global solutions and overlay networks in particular when thinking about these things. I actually like the idea of an overlay network and I've thought about doing it myself, but I can't really convince myself that it's more practical than things like sofiechan. But maybe there's a decentralized overlay network in sofiechan's future.

The problem I find more interesting though is the problem of the eternal september. In particular, it's called september because the internet used to be dominated by college students, which were a fairly comprehensive selection of the top k% of first world society. That's a damn fine selection right there, and there's a reason it's still remembered as the ideal. So in all this self-solving self-selection into semi-exclusive social networks, is there ever going to emerge again a candidate for *the* social network for the broader elite-adjacent class of a particular society. Facebook started as this but quickly decided to connect everyone in the world instead of just college students. Could it be done again?

I think it could be done again. I think if you took a large public social network and applied a selective entry filter that more or less any real high quality person could pass (ie prove that you are a college student or grad, pass IQ test, get vouched in, or otherwise post proof of physis), you get something quite powerful. Who doesn't want to be a member of the network that includes only the best, if it is in fact good? Millions of people pay hundreds of thousands of dollars for this! Such a network would become an institution of the elite, if it worked at all.

I think there's a lot of value in doing that, whether decentralized or not.

referenced by: >>3587

As the internet beco received

anon_voge said in #3587 7d ago: received

>>3577
>>3581
While I agree that sofiechan seems to attract high quality members, there is nothing preventing a million Indians finding the site tomorrow, filling it with slop, and killing it.

I am thinking bigger than just a forum or social media platform. What if I want an email address that only my peers can physically reach? Maybe I want to sell a service or host some media that only a subset of the globally diverse internet can access? These are questions that a general-purpose network solve. I am imagining something more like Urbit, without the retarded architecture.

>Who doesn't want to be a member of the network that includes only the best, if it is in fact good? Millions of people pay hundreds of thousands of dollars for this! Such a network would become an institution of the elite, if it worked at all.
Exactly. But I suppose attracting those initial best is the hardest part.

referenced by: >>3588

While I agree that s received

admin said in #3588 7d ago: received

>>3587
Ignoring the racism and speaking strictly to the technical point of sofiechan's defenses against a large influx of low quality noobs, the voting system here is entirely designed around that eventuality. Specifically it is designed to enable the incumbent community to effectively filter out low quality noobs, prevent them from gaining a foothold, and give the admin and core community even ability to fight a successful reconquest. Almost every parameter in the system right now is tuned to be absurdly egalitarian, but the instant that that becomes a problem this place can become extremely elitist and automatically identify and purge the low quality masses in about 15 minutes. This is not very tested, but I am confident we would be able to weather plausible virality scenarios with quality intact. And I will be continuing to think ahead and build the ultimate taste machine.

>But I suppose attracting those initial best is the hardest part.
So far so good.

Ignoring the racism received

anon_tira said in #3589 7d ago: received

What's the post-mortem on Urbit? The new owner recently crashed out on Twitter with a long essay I didn't bother to read. It seems like to me like the concept of Urbit had a lot of potential but there was something profoundly wrong with how it was being implemented. I used it between 2019-2021 and within that entire timeframe they couldn't get the messaging app to stop crashing constantly. They did attract high-quality people but it definitely didn't suffice to get the thing off the ground.

referenced by: >>3590 >>3591 >>3592

What's the post-mort received

anon_voge said in #3590 7d ago: received

>>3589
I’m in the same boat. Used it for a while, but got fed up with the buggy software. Not sure what they’re doing now.

The main issue I always saw with it was that the idea was to build _everything_ from the ground up, instead of using tried and tested protocols + software. I love the buy in private p2p network aspect, less so everything else.

referenced by: >>3591

I’m in the same boat received

anon_betu said in #3591 7d ago: received

>>3589
>>3590
Yeah I've got an urbit star and I really like aspects of the project, but building everything on a novel computing formalism, OS, ground-up overlay network, etc was ultimately the wrong choice. Even a novel protocol and software could have been warranted, but Curtis delved too deep.

The other big problem with Urbit was that the killer app thesis was never substantiated. Ideas like shrubbery and this anonymous discussion thing Curtis is talking about now are cool, but I don't see what it needs Nock and such for.

I think an urbit-like identity layer and overlay network protocol plus CRDT datastructure sharing possibly with not-quite-turing-complete scripting learning from the success and failure of javascript could be very cool, but you basically rip out the whole notion of trying to control the substrate. It's got to be about what language you speak, not what code you run.

referenced by: >>3602

Yeah I've got an urb received

anon_kaxy said in #3592 7d ago: received

>>3589

Urbit is a prime illustration of the limits of the theorycel.

Curtis Yarvin is one of the all-time great theorycels and we love him for that. A gadly, a provocateur, a lively conversationalist and a vivid writer. There are domains where this disposition works well, like posting. This is not snark--I say this with real respect--theorycels are good at expanding imaginations, which is a critical function in a society that wishes not to stagnate.

But there are many other domains where you need pragmatists. Politics, for example. See this political strategy debate between Yarvin and Rufo last year, an exchange that was great, informative, and also a brutal first-round knockout for our theorycel friend:

- 20 second version: https://x.com/RushDoshi/status/1882534504440660155
- 20 minute version: https://im1776.com/2024/04/11/rufo-vs-yarvin/

Anyway, working software comes specifically and only from pragmatists. It is a domain where managing complexity is a paramount concern and choosing boring-but-correct tradeoffs, a necessity. "No plan survives first contact with the enemy"--no software survives contact with real users. Success depends on a long ground campaign of such boring pedestrian things as "descoping", "refactoring", and "fixing bugs". Great software, like all great engineering, is ambitious in a single dimension towards a well-defined goal, choosing known-good standard foundations for everything else.

Yarvin, then known as Moldbug, started Urbit more than two decades ago with a grand vision to redefine software, networking, and more into a unified cathedral of computer engineering. A new programming language, "Hoon". A new assembly language, "Nock". A lot of other new things with what the man himself might call Star Wars names.

Urbit is just one in a long history of this genre of utopian maximalist software project built by a charismatic artist-type. There was Xanadu in the 1960s(!), a top-down pre-world-wide-web hypertext system which Ted Nelson is still technically working on 50 years later at 88 years of age. Then there was MojoNation in the 1990s: "what if we rebuilt the everything from Scratch as a Decentralized Peer-To-Peer Protocol." That story has a happier ending. Bram Cohen quit MojoNation and extracted one small part of it, the file-sharing part, calling it BitTorrent.

This is the best outcome Urbit can hope for: that a pragmatist might extract a good idea from its thicket and use it to advance the plot. I'll offer this: I understand at least Urbit's ID system (~faldyr-silrux, etc) in detail; it's open and extensible, built on Ethereum. If anyone has the will and the way to build something real in the "~" cinematic universe, I am happy to support and advise.

TLDR; Moldbug is a talented shitposter and a bad technologist. Urbit is a great parable covering art, engineering, our individual natures and the exhortation of Delphi to "Know Thyself".

referenced by: >>3601 >>3602

Urbit is a prime ill received

anon_wuro said in #3601 7d ago: received

>>3592

Bravo, anon.

Bravo, anon. received

anon_voge said in #3602 6d ago: received

>>3591
>an urbit-like identity layer and overlay network protocol
This is exactly what I envision. Just TCP behind some kind of access control. That way anyone on the network can start hosting and using the software they know and love, and building on top of it in ways they know how. In my opinion Urbit has failed to thrive (or was even DOA) due to them literally reinventing the wheel using the most obscure and convoluted tech possible.

>>3592
Well said. Urbit is a grand vision, whether it lives or dies. It's just too abstract I feel.

This is exactly what received

anon_mizi said in #3605 6d ago: received

>>3574
I could absolutely imagine this being something real, but as others have pointed out in the Urbit case, architecture astronautics like implementing a Yggdrasil style overlay network must absolutely be entirely abstracted away from the people actually using the platform. The problem is that it's practically impossible to engage in serious architecture astronautics these days without having a tangible, negative impact on your users. It's not 1969 anymore. Even worse, it gets harder every day to fly away, because the inexorable march of the incumbents continues while you sleep.

If you want your paid private internet to be more than an insanely glorified groupchat, the way in has to be gated through something that is as normally accessible and frictionless as a login page in the browser and a stripe checkout page. If you have to, for instance, touch the command line, history says you're pretty well fucked. And if your stuff crashes after 20 years of Kelvin versioning because it's still running Martian bytecode at the end of the day, you're also pretty well fucked.

I could absolutely i received

anon_gyvo said in #3610 6d ago: received

Until there's a particularly elegant technical solution one can always fall back on defense-in-depth. It's great that this forum has a nuclear option for handling an influx of noobs or raids or what have you, but the form keeps most of the wrong sort of poster out.

Anonymity means it's hard to chase clout. Minimal images means less virality. Posts being by default long and in complete sentences filters out anyone with a tiktok attention span. The content is heady and often obscure, etc.

(Of course it will always be tempting to reverse all of those things for reach.)

I think it's worth designing as much for uninterestingness and unintelligibility as for technical defensibility, both in technical form and cultural norms.

Until there's a part received

You must login to post.