Can Tlon start bringing more of its engineering discussion into this public forum?

108 views
Skip to first unread message

Christopher King

unread,
Sep 20, 2020, 4:39:20 PMSep 20
to urbit-dev
I could totally be speaking out of turn here--perhaps most of the engineering discussion is here already. But a few things I've read recently do seem to indicate that quite a bit of discussion about the future of Tlon's efforts to build out Urbit's infrastructure seems to take place in private Tlon settings. This likely was not really an issue thus far, but I'd like to encourage you as an outsider to open up a little bit more.

Given Tlon's role as a nurturing presence for a decentralized platform far beyond its borders, it's probably worth getting outside feedback for engineering decisions, or at least setting up a process (i.e. publicly visible threads) for such feedback when someone wants to provide it. I am absolutely sure none of it is intentional and it's probably a by-product of just being super focused on what you're trying to deliver, but in the process of creating my hosting platform in the last few months there have been a few different occasions when it felt Tlon being a little more open could have been helpful. For example, I had no idea until recently that Tlon was also working on a hosting solution. (It is certainly a possibility it was announced before the 8/18 blog post and I just missed it.) And when I've tried to make Github PRs, it hasn't felt like the most accommodating atmosphere sometimes because the future plans for a given section of the codebase aren't visible to outsiders.

In any case, just some food for thought. I always try to leave the feedback I'd want if I was in someone else's shoes. I could be wrong and the balance you're currently striking could be the right one for your objectives.

Matilde Park

unread,
Sep 20, 2020, 4:48:55 PMSep 20
to Christopher King, urbit-dev
It’s not out of turn, don’t worry. I totally agree. We have been making a distinct effort to do so since around August 25th 2020, actually. And I think we haven’t quite had any engineering discussions specifically *internally* since then.


~haddef-sigwen
https://urbit.org

Galen Wolfe-Pauly

unread,
Sep 20, 2020, 11:41:11 PMSep 20
to Matilde Park, Christopher King, urbit-dev
In general: yes. Most emphatically with regard to kernel and runtime development.

For the most part, the opacity is due to exactly what you mention: we're just busy trying to make this thing work well. The decision to split out urbit.org is the first step toward making the kernel and runtime as open as possible. If there's anything you're curious about, we're happy to answer questions and help. 

We've talked casually about offering a hosting product here and there, but I don't think we've ever messaged about it publicly. My hope is that many Urbit hosts come to exist — not unlike the variety of hosts you might find for running a conventional web server. Or maybe it's more similar to a WordPress host — but you get the idea.

Christopher King

unread,
Sep 21, 2020, 9:21:23 AMSep 21
to Galen Wolfe-Pauly, Matilde Park, urbit-dev
As far as the hosting product goes (and I guess I'm hijacking my own thread here), I also encourage Tlon to think carefully about the extent to which any services it offers will automatically be seen as the default. I don't expect deference to my platform, of course, but a big impetus behind me building this out was a desire to capture network effects. I know if I was in the market for a hosting solution, absent a significant price differential, I'd probably want to go with the seemingly "official" Tlon solution rather than a third party. This goes beyond just a hosting solution, but any sort of Urbit products on offer. Tlon's offerings will to some extent build out the network, but they may also cannibalize third-party infrastructure, which itself probably harms the network (in as much as the motive for non-Tlon builders will be diminished).

The above I speak as objectively as possible. From a more self-interested perspective, once sufficient consensus within Tlon is reached to officially list my solution, would it be possible for Tlon to market its own solution as not an "official" one? Maybe brand it with a different name, so that it appears "equals" with my solution. In that case, Tlon will have the option to delist my solution if it does not perform up to par, but it might avoid inadvertent cannibalization that harm the network. Of course, if Tlon is counting on the financial income from hosting, that's a different matter entirely and the above may not represent Tlon's true perspective and interests.

Matilde Park

unread,
Sep 21, 2020, 9:24:03 AMSep 21
to Christopher King, Galen Wolfe-Pauly, urbit-dev
Well the burgeoning Urbit Foundation has a pledge to all developers, not just Tlon. We don’t see Tlon’s output as official, just flagship. And thus of course they will be seen as equals. It’s not Tlon plus the world — Tlon is just a corporate bootstrap getting Urbit started, and urbit.org is loyal to all parties working around the project. 

On Sep 21, 2020, at 9:21 AM, Christopher King <ch...@cking.me> wrote:



Christopher King

unread,
Sep 21, 2020, 9:31:24 AMSep 21
to Matilde Park, Galen Wolfe-Pauly, urbit-dev
I understand you don't see Tlon's output as official, but users might. If I was a new user, I would not distinguish between official and flagship.

Tyler Shuster

unread,
Sep 21, 2020, 10:24:13 AMSep 21
to Christopher King, Matilde Park, Galen Wolfe-Pauly, urbit-dev
There’s a lot of momentum associating tlon with urbit but I think the recent movement unassociating urbit.org with tlon will help a lot. If you just went there today and looked through, it would only suggest the primacy of tlon, not its supremacy.

If urbit’s software linked to tlon that would be a problem, but it doesn’t. At this point the only place I really see tlon as formally in charge of urbit is in the GitHub repo, and I do think the Foundation has some work to do in establishing some formal processes around that. But as I said, momentum is strong. I agree with Christopher about opening up more processes and I’m pleased with the steps tlon has taken so far. 

T

On Sep 21, 2020, at 6:31 AM, Christopher King <ch...@cking.me> wrote:


--
To unsubscribe from this group and stop receiving emails from it, send an email to dev+uns...@urbit.org.

Galen Wolfe-Pauly

unread,
Sep 21, 2020, 11:34:23 AMSep 21
to Tyler Shuster, Christopher King, Matilde Park, urbit-dev
There is *so much* room for providers to differentiate themselves. Price, security guarantees, performance, custom software, quality of support, user experience. No matter what Tlon does, there will be other niches to fill. 

Josh Lehman

unread,
Sep 21, 2020, 11:48:59 AMSep 21
to urbit-dev, ~radbur-sivmus, Matilde Park, ga...@tlon.io, urbit-dev, Christopher King
Read you loud and clear Chris. We're in the early stages of opening up development to the community more intentionally and really appreciate the feedback.

Regarding hosting, we'll do our best to ensure that urbit.org (the website) is not opinionated about hosting providers. To that end, it would be great to present https://www.geturbitid.com/ on our public documentation. I can think of one place off the top of my head where this should go: https://urbit.org/using/install/ > Setting up a planet > Choose to host or run your ship locally

If you submit a PR to the urbit.org repo, we'd be happy to get it listed. When Tlon launches its own hosting service, we'll list it there as well, I'd think.

Jon "Poprox" Paprocki

unread,
Sep 21, 2020, 12:36:32 PMSep 21
to Josh Lehman, urbit-dev, ~radbur-sivmus, Matilde Park, ga...@tlon.io, Christopher King
On Mon, Sep 21, 2020 at 11:49 AM Josh Lehman <jo...@urbit.org> wrote:
Read you loud and clear Chris. We're in the early stages of opening up development to the community more intentionally and really appreciate the feedback.

Regarding hosting, we'll do our best to ensure that urbit.org (the website) is not opinionated about hosting providers. To that end, it would be great to present https://www.geturbitid.com/ on our public documentation. I can think of one place off the top of my head where this should go: https://urbit.org/using/install/ > Setting up a planet > Choose to host or run your ship locally

If you submit a PR to the urbit.org repo, we'd be happy to get it listed. When Tlon launches its own hosting service, we'll list it there as well, I'd think.

There is already a discussion about this ongoing in the main geturbitid thread. I'm just waiting for the thumbs up from the Tlon people currently testing before listing.

Josh Lehman

unread,
Sep 21, 2020, 12:40:30 PMSep 21
to Paprocki, urbit-dev, ~radbur-sivmus, Matilde Park, ga...@tlon.io, Christopher King
Ah, great—thanks for the context!


~wolref-podlex

Tyler Shuster

unread,
Sep 23, 2020, 12:09:43 PMSep 23
to urbit-dev
As I’m watching the UC/Development chatter this morning, this thread comes to mind. Four observations and a proposal:

I’ve seen in some screen captures and heard conversations about the Tlon group on Urbit, which is where I hear a bit of conversation about Urbit development goes on.
I’ve participated in a bit of conversation on GitHub about Urbit design and development.
The Design/Development channels on UC seem mostly relegated to speculation, and mostly not for those of us interested in actually doing the work of designing and developing Urbit. This is not a judgement, it’s just what I’ve noticed those channels have become.
This channel (urbit-dev mailing list) seems to be the most active channel for actually discussing these things, but as we’ve implicitly noted this morning, getting notifications whenever there’s activity can be a bit tiresome if it’s not on a subject that we’re interested in.

Therefore, as this community seeks to delineate the roles of Tlon versus the larger Urbit community in developing Urbit, I wonder if we could have a group on Urbit specifically focused on doing the work of building the system, that’s open to Tlonners and nonners alike. We could move a lot of discussions that happen on that channel to there, and we could keep the various discussion threads more focused. I would also love to see a general “Urbit promotion” channel for those of us who are trying to do the non-software work of propagating information about Urbit. For example, some chats:
Web Interface Design
Command-Line Interface Design (?)
System Design
Hoon Development
Nock Development
Urbit Promotion

This does get a little muddy when it comes to Landscape development, which AFAIK is a Tlon product that just happens to be bundled with Urbit due to the nascency of the system. 

I’ve had some long threads on Github talking about these subjects and I cringe a bit when they turn into discussions about possible designs because I’d rather have those discussions on Urbit a) because dogfooding and b) because they should be open to the community (not everyone is on Github)
As Publish gets more refined, we could also move long-form discussions like we’ve had on here into that
Overall, I think it would be a good move to consolidate some of the discussions that are happening on the four disparate channels (internal boards, Github, the high-noise UC channels, and this mailing list) into our own software. Of course, you run into the classic “now there are 13 competing standards” dilemma, but I think it could be done right.

Matilde Park

unread,
Sep 23, 2020, 12:30:08 PMSep 23
to Tyler Shuster, urbit-dev
> I’ve seen in some screen captures and heard conversations about the Tlon group on Urbit, which is where I hear a bit of conversation about Urbit development goes on.

The Tlon group is literally just internal stuff, not really as Urbit project / Arvo development oriented as you’d think. In the same way that we do have internal mailing lists for “out of office” notices or the like. Sure, stuff like Tlon hosting has only internal chats, but that isn’t Urbit development.

> Therefore, as this community seeks to delineate the roles of Tlon versus the larger Urbit community in developing Urbit, I wonder if we could have a group on Urbit specifically focused on doing the work of building the system, that’s open to Tlonners and nonners alike.

This is worth thinking about; I have heard ideas thrown around for groups for Urbit.org stuff separate from the Tlon-internal projects, and I think with the new groups layout in Landscape showing all channels in one sidebar, and notifications landing soon that make channels opt-in, it would make it far friendlier for splitting development channels as such.


~haddef-sigwen
https://urbit.org

Christopher King

unread,
Sep 23, 2020, 2:35:14 PMSep 23
to urbit-dev, urbit-dev
I do think it makes sense to migrate such discussions to Urbit itself to the extent possible. Tyler lays out a pretty thoughtful and explicit plan that I think serves as a good roadmap.

As a nonner, I do think there are handful of things to be thoughtful about while doing so, though. Splitting things out should not just serve the purpose of making it easier for some Tlonners to ignore. Tlon is Urbit right now, whether it wants to be or not. It controls urbit.org (for now), approves all contributions to the Github repo, and employs the majority of the engineers worldwide that know anything about Hoon and Urbit's plumbing. Nonners generally have no idea who at Tlon knows about what unless they've personally answered a question for you on the subject previously. Especially at this stage of Urbit development, when it feels like the ground is ripe for an exponential slope of adoption, Tlon should explicitly adopt more of a "customer service" mindset. Generally I've been deeply impressed with the willingness of Tlon engineers to support a nonner and answer questions. However, some level of annoyance clearly arose today at having a pesky outsider constantly pollute urbit-dev and "promote [his] product", the self-serving bastard. We're not competitors here, and strengthening or building out the network helps all involved.

In general, I just think it's important to continue the process of making Tlon more open, which I think has been going great so far.

Galen Wolfe-Pauly

unread,
Sep 23, 2020, 5:20:50 PMSep 23
to Christopher King, urbit-dev
> I wonder if we could have a group on Urbit specifically focused on doing the work of building the system, that’s open to Tlonners and nonners alike
This really is the idea behind Urbit Community. I think pretty much everyone from Tlon is there. I could definitely see us adding channels and expanding into other topics. As the groups experience gets better it'll also get easier to engage with other content types (publish posts, links). Right now Landscape is pretty chat-centric.

> Overall, I think it would be a good move to consolidate some of the discussions that are happening on the four disparate channels (internal boards, Github, the high-noise UC channels, and this mailing list) into our own software.
Agree completely. In some ways, this is the primary driving force behind Landscape. 

There's generally always going to be tension between design work, which is often best done with a small number of people in a private form, and open source work, which leverages public participation and contribution. We couldn't possibly do all of our design work in public, but we can probably do more of our development work in public and clarify what we expect to deliver and when. As discussed, we're working on that. 

There's always going to be a difference between urbit.org and Tlon. The former, by default, will always be more open than the latter. It's true that Tlon has always been the caretaker of urbit.org — but that's changing, since we're clearly at the point where the system has to be built in public. Landscape does exist as a strange grey area. It's basically Tlon's best attempt at making something useful on top of Urbit. We want it to be open source, but we also want it to be a great product. These things are always going to be in tension. It's easy to imagine a distinction similar to Chrome / Chromium in Landscape's future. Whereas I expect Arvo to, as much as possible, be like the Linux kernel.

> Generally I've been deeply impressed with the willingness of Tlon engineers to support a nonner and answer questions. However, some level of annoyance clearly arose today at having a pesky outsider constantly pollute urbit-dev and "promote [his] product", the self-serving bastard. We're not competitors here, and strengthening or building out the network helps all involved.
We do want to help you. I think there's a difference between providing support and announcing a new product. Probably better not to use this list as a support channel, although the technical threads you're starting are useful to have for posterity.

Ed

unread,
Sep 24, 2020, 1:29:45 PMSep 24
to Galen Wolfe-Pauly, Christopher King, urbit-dev
> Therefore, as this community seeks to delineate the roles of Tlon versus the larger Urbit community in developing Urbit, I wonder if we could have a group on Urbit specifically focused on doing the work of building the system, that’s open to Tlonners and nonners alike.

It's funny this thread is coming up when it is — just yesterday, I was thinking we've reached a point where we should seriously consider creating an "Urbit Foundation" group, for this exact purpose.

~fabled-faster

Christopher King

unread,
Sep 25, 2020, 8:46:04 AMSep 25
to urbit-dev
Jeez dude.

On Fri, Sep 25, 2020 at 4:10 AM Jake Miller <ry...@hey.com> wrote:
First, a disclaimer - I don't work for tlon, I've never really been part of any inner circle (except for maybe just sticking around for a while), and while I've had my issues with OTAs breaking my shit, urbit is open enough to figure stuff out for yourself. I've gotten pretty far with that. I've never taken a hoon course (though technically i signed up and just ended up being a guide), I've spent hours and hours figuring out esoteric code that changed the next day, and my interaction with tlon engineers at a deep technical level has been /relatively/ non-existent, but mostly because i just like doing things myself. Yet I have never felt any semblance of elitism or unwillingness to be as open as possible - in fact, in both urbit-dev and many of the channels, I've seen a consistent and compulsive effort to be incredibly transparent on what is going on. ~palfun made a change recently that broke srrs, and i figured it out, mentioned it, and a couple of hours later had a response from him about making sure to announce breaking changes. Sure enough, there is a breaking change section in the OTA emails now. I've made plenty of PRs that have gotten immediate responses, and with so many 'nonners' contributing i don't see how anyone could get the impression that tlon is being anything less than what they're claiming.

Secondly, just as an example of how much space there is for hosting providers - I'm working on a project that allows you to associate and authenticate gaming profiles with your Urbit ID. One of the components in making this happen will most likely require a 'server ship', run by the host of a particular game, that will at first play the role of a login server. This alone already creates a clear demarcation between a vanilla hosting service - they might just want a pre-configured server ship. Even if they are willing to set one up themselves, if you're in the business of blindly hosting anyone who can pay up, there is a good chance that a company looking to build a game supporting Urbit IDs may not want to use your services, and players may find that buying cheap planets might prevent them from playing their favorite game. Companies can't afford to purchase a planet from a provider that may be selling planets cheaply in bulk to gold sellers (then you're stuck trying to switch sponsors who might not want you or just back to banning ips, which we know doesn't work, and becomes a dealbreaker for players). They may even invest in a full star, which brings up a whole new differentiating factor as an avenue for crowdfunding games. People love getting "Founder" rewards - what if they were etched into your digital identity, a status symbol that no one can ever take from you? And you get an OS, not just an ephemeral promise that maybe you'll get the thing? 

And de-incentives for botting and gold scammers is the tip of the iceberg. Private torrent groups with server owned moons anonymizing traffic and auditing torrent activity, location based ala nextdoor but with the ability to curate your local feed, hell maybe someone will build a salesforce / insperity to sell as an HR solution for enterprise. I specifically want to see an education based provider that contracts with schools, because what kids are dealing with right now is borderline abusive, failing because the systems are so inflexible and buggy that half your answers could be erased after submitting a test, and there's absolutely nothing you can do about it. 

I don't think you could find a clearer demonstration of how the sponsorship hierarchy can become the foundational aspect of an emergent and clearly enforceable reputation based system.This won't be an AWS vs GoDaddy vs DO - providers will evolve according to the needs of various communities, the reputational level at which they're willing to operate, and the extent to which their handling of bad actors maintains that reputation. A provider focused on gaming could provide anti-bot support through a scalable set of moons that communicate with the game's server ship (an agent is a user programmable database, no?), look for suspicious player activity, maintain a list of botters/scammers/, and send that information back in a form that is agnostic to the specific game. Can't afford that? Fine, find a cheaper provider. Maybe they do those things,  but just a bit worse. Both hosting and planet providers will have to do deal with an incredibly fragmented landscape, and that's to their benefit. I wouldn't even be surprised if tlon's hosting becomes the linux desktop of the provider ecosystem, with the less techy preferring whatever caters most to their needs.

Last point, and apologies if this comes off dickish. This complaint that tlon has an unfair hosting position reads like the government just built a bunch of subsidized housing and they have an advantage because people trust the government and won't buy my slightly cheaper and equally boring housing. This misses the point - urbit is about building, it's anti slumlord, and complaints centering on name recognition and assertions about what tlon's proper role should be invoke the same 'ugh field' as a 15 year old jumping into a mcdonald's ball pit right when you get to that perfect depth. You have land. Great. But it's Kansas farm land with no particular advantage over the 800 acre ranch 50 miles from you, and if you're annoyed that they've decided to parcel it up and sell it I think you're losing sight of what problem urbit is trying to solve. I'm sure we all would like it to grow, but we don't want a Titusville and god forbid if we ever get an Ida Tarbell. 

Hit me up at ~littel-wolfur if you're interested in discussing this, or just want to call me an asshole. But this whole thing sounds like bitterness wrapped in a complaint that might've been relevant 5 months ago, but certainly has no merit now. It's a hosting service - anyone with a star has probably thought about making a quick buck and selling some planets, and while useful, probably the most uninspired idea you could come up with w.r.t what you could do with your stars. There have been many discussions on opening up the development process, but just tracking commits on github gives you 99% of what you're after.

Urbit has been my happy place for a while now, and a lot of it is because everyone builds stuff and doesn't complain. I don't want to see tlon's diplomatic attitude towards criticisms like this turn into bitterness towards users. And just to offset all this white knighting - tell me how to make fucking dynamic tiles again! i miss seeing review counts on srrs :'(

Matilde Park

unread,
Sep 25, 2020, 9:14:19 AMSep 25
to Christopher King, urbit-dev
Just a touch harsh, Jake. :) Civility!

I can accept one would feel a little spurned by how private we’ve been about hosting. 

We really are trying to be clearer with our userspace developers, though — someone once mentioned to me that a year ago the concept didn’t quite exist and our standards are changing in regards to explaining above and beyond code changes, getting ahead of the deploy so everyone can be prepared. Versioning APIs. The upcoming Landscape dev stream, so we can cycle a ‘stable’ release for the first time. You get the idea. 

On Sep 25, 2020, at 8:46 AM, Christopher King <ch...@cking.me> wrote:



Christopher King

unread,
Sep 25, 2020, 9:44:02 AMSep 25
to urbit-dev, urbit-dev
Oh, I don't feel at all spurned. I'm a big boy. One of the main impetuses behind me taking my participation in Urbit in this direction is that, in the final estimation, it does not rely on any sort of decision from Tlon or its affiliates, although I of course do hope for urbit.org listing and generally being friendly with one another.

My approach, informed by what I've seen to be the best approach at companies I've worked at, was just to provide constructive feedback from the outside. I firmly believe, as a principle of business, that the recipient basically can't have enough of that, as long as it is given in measured terms and in good faith, which it was. It has a great monetary value, in fact, as companies regularly spend large sums to divine what their users/customers are thinking. None of it has to lead to any change, it just represents one user's experience. I thought my surprise at the impending release of Tlon's hosting solution was worth communicating for this purpose. It certainly does not preclude the possibility of my not having fully investigated all streams of communication.

Anyway, just trying to help. But so was Karl Marx, I guess. ¯\_(ツ)_/¯

ryjm

unread,
Sep 25, 2020, 1:27:20 PMSep 25
to Christopher King, urbit-dev
Re-read this in the morning and I did not realize how harsh those last couple paragraphs look. Swear I’m not usually like this, guess the day job is getting to me. I clearly read too much into your suggestions / experience and chose the most pessimistic interpretation. Maybe I'm the 15 year old in the ball pit! Ironic.

So yeah, feel pretty bad causing what might amount to the first unfriendly conversation on urbit-dev, so please let me know if I can help out in any way - as a fellow nonner I can appreciate the frustration, but i also know my way around now if someone from tlon isn’t available. I was thinking of starting my own hosting solution for the gaming stuff I talked about, but maybe we can partner up on it a few months down the road if you're up for it.

The friendliness and openness of everyone here is one of my favorite parts about urbit, and i’d hate to be evidence of the opposite. Sincere apologies for coming off like a jerk.

Galen Wolfe-Pauly

unread,
Sep 25, 2020, 1:30:27 PMSep 25
to ryjm, Christopher King, urbit-dev
It's funny, I was going to write something about how it's probably okay for us to see a bit of conflict around here. 

I basically expect that everyone here wants Urbit to succeed, and to be better. There's no reason to be shy or indirect — quite the opposite. It's fine to go at it, so long as we can stay civil. 

Anyway, I think both perspectives here have merit. We (Tlon) can be clearer in certain ways. And, the ethos of Urbit has always been that you can read the source to figure out what's happening. 

Tyler Shuster

unread,
Sep 25, 2020, 2:16:15 PMSep 25
to Ed, Galen Wolfe-Pauly, Christopher King, urbit-dev
I like this idea of an Urbit Foundation group. Here’s another thing that’s been on my mind:

When new urbiters spawn, it’s nice to have a friendly place populated with townspeople willing to help you learn the ropes.
But it feels like the Design and Development channels are part of that same spawn point, so trying to do work there feels like it gets crowded out by people who just want to talk about any kind of design or development — newbies and old hands alike. 
And I think this is a function of the Urbit Community’s status as “Spawn point town square.”

Watercooler is great, urbit-help is great, Meetups & Events is good, I guess “Outdoors” and “Hosting Party” exist…? didn’t know that until I had to rejoin all the channels this morning. But the others, I think both a) make it hard to have a focused place to do urbit-related work and b) contribute to the “Urbit is a clean-slate OS for talking about urbit” meme, which happens when the default location is also the development forum.

So I like the idea of Urbit Foundation — it certainly feels like a different group of people and it might ameliorate these issues.

Ed

unread,
Sep 25, 2020, 2:29:32 PMSep 25
to Tyler Shuster, Galen Wolfe-Pauly, Christopher King, urbit-dev
Yeah, I second a lot of these thoughts re: setting up an Urbit Foundation Group, for explicit production/work/work discussion purposes. 

In this scheme, with a "friendly" Urbit Community and a "work oriented" Urbit Foundation groups, we'd have our spawn point town square, and the nearby workshop from which major works are furthered, developed, proposed, argued, etc.

~fabled-faster

Reply all
Reply to author
Forward
0 new messages