Venkatesh Rao: And why don't? Get started with the intro and we can post a note for late joiners to catch up and i've started the transcription so. That record will also be there.
Mike Travers: Okay. Okay. Okay. to share this.
Venkatesh Rao: .
Mike Travers: Are you? Are you seeing this? so so, ? Hi, everyone, , good morning. , so Im going to go over? These. I apologize because this is. It's not very. Coherent, it's kind of a kind of a messive, a massive ideas, and it's not not very, not very formal for a formalism group, but I. I hope there. Some interesting, interesting pointers in here and. And I'm just. Kind of? Kind through all this stuff and feel free to ask questions or interrupt at any time. So act. So actor? So actors are a bottle of concurrent of concurrent community community, concurrent computation as intercommunicating objects. For first develop. For carl, He. Hewitt in the early early 70 S. And it kind of grew up in parallel with with , the more better known ideas of objectoriented programming, I'll talk a little bit about that. . . it's It's both. . Trying to be both, , , so. A very abstract formal model for computation and a model for real world, distributed systems, which I think is what makes it interesting? There's not really a single formalism. it's more of a conceptual model. there's a number of different. You know practical implementations and languages that use it, I'll. Point to a few of those to. . . and I don't know if it's really that interesting except for his historical reasons, , you know I don't it's? it never. It's not quite a P. Paradigm, . But it's I think it's historically interesting, and it's close to object oriented programming and agent based modeling, which I think are very relevant to, to protocol modeling so.
Venkatesh Rao: .
Mike Travers: So I'll Try and tie those things together. , So actors themselves Are you know extremely simple models so. As a kind of a mathematical formalism that that's good. . . basically, and and? An actors, you know. Do? Do? things while they, they're An object, they, they receive a message from another, another actor. , they. , generate messages of their own to to other actors, . . . . sorry. . sorry. here, ? That's , So they. Change their in. Internal state, which, because? Conceptualizes changing their behavior so they have a behavior time t once they get a message. They can replace that with a new behavior at time, T one, that's a way to represent internal state. They send out other messages and they can create create new actors. That's that's actually. Fairly powerful. Function thats not found in a lot of other formalisms, .
Venkatesh Rao: .
Mike Travers: So? So? So? ex. Extremely simple. . . of. The way this grew up was also what at? At Mit, while people were trying to figure out, actually. The practical versions of the Lambda calculus, So if if people are familiar with the The Scheme language, which is? Popular, you know, teaching dialect of list, but teaching language, . Actors and scheme kind of grew up very in parallel, and their, their, almost. they're almost the same thing, except that actors is more focused on concurrency than Scheme. . . . what do we have here so that thats about it. Here, I have a link to a video of of Carl Hewett explaining actors, which, if you might, M. Might be interesting, Carl. Carl here was a. , and? A strange guy, , And and he's It's worth getting getting you know hearing what his thought processes are like. . . again. I'm not going to dive too much into the the formal details of this but a. Sound a lot like object oriented programming which most people are are familiar with. That's not a coincidence because they, these ideas kind of were developed, developed or at the same time, there's a lot of interchange between. Hewitt's group, and and. Allen Kay's group at Park, which was developed developing small talk and pop. Opportunited programming, . , I. , really recommend. This this paper on the early history of small talk by by Alan Kay. , Which? Goes into. And the reason I think this is this is interesting? Is that Obj? Oriented programming I view, sort of like the the , protocolized view of computation, So you? you see these people you know, the very you know you know what are what are now very early stages of the development of of computer technology, you know, trying to think about computation. In protocolish terms of of. Separate things talk talking to each other, how do you organize that? What? what does it mean? I, I find the intellectual history kind of fascinating, even though this, this stuff is more now, I think kind of part of part of common knowledge, it wasn't wasn't back then. . . . see? . . this is skipping around kind of randomly so. So so? Swit. Swit. to so some of the the formal computational issues in actors and im not going to. Dive into these, but ? But But But to you know, actually the actual formalism, so actors are inherently async asynchronous. , you? But bunch of bunches of objects, they're sending messages to each other. there's. No particular guarantee of what the message will get through, or what order events happen in , so that means there has to be buffer buff. Of messages, and there has to be, and because buffers can't be infinite, there has to be some model. Back pressure of trying to. Have have some control over the. The flow of of messages through through the system. and act. System was radically nondeterminism because. Because of this distributed nature. There's questions about fairness and guarantee of delivery and. Actor about systems tend to be very fine grade, So so the the other distinctive feature of actor systems is how they handle state. . . , and? And so? So state? So than being in a central database, every actor kind of maintains its own its own state, . In the actor model these these states. Tends to fairly fine grained, and so an active might, but a typical might a bank account that keeps up, but keeps a balance. for instance, that's one of the typical examples. Think of examples. here's , here's another. Here's another. Kind of manifestation of actors? . . their use in animation, so one of the. . One. One. . Semi-famous early example of computer animation which I wont because I won't play here was. Done done by Craig Reynolds, who did a model of Flo? Flocking behavior of how so? How large groups of? Of actors. Coordinate their movements, . This was this was done on. List machines, it was then it was done done using and an actor model again. This pretty pretty crude. By today's standards, I think we'll play a little bit of this, ? Pretty crude by todays standards, but. Advanced first time in the. The. The. was was used to make the animation system. And? And? flocky, the flocky algorithm. Flocking algorithm has. Pretty influential in just elsewhere, so here you can see. This case. This bunch of stuff fish. The. The. model. The. model. Model here here is simple, its its sort of a physics, , physicsy model. Where each, , each bird. Has a radius of other birds that it pays attention to and they try to orient their. Their their? Their their. Their their. that. Very very simple model, very powerful, and and and this was so, this was an early early kind of actor success story, And if you? , . , so? And? so so? And? So? So? the actor model itself. Has it has? Has the problem that. That it's it's really much to. I think much to computer science theoretical, So I was, I was looking for I was, I was trying to look for concrete. Examples of complicated actor systems who have been involve. You know lots of different act. Doing communication and then there. There were surprisingly hard to hard to find in the. The academic literature, all the examples of actor languages are doing. you know extremely simple things like factorials or or bank accounts, because I don't know Because of that's out, theoretical computer science goes. However, actors did have some exten. Into real systems, aside from sort of the animation, so earlo? An actor based language that actually has been used. Industrial strength applications in in telecommunications and other other things. . . . . think it's Eric said. Is the company that that? Popularized our line of those applications and have interesting that. Not only so. So, , so. Phone call generates and generates an actor, which manages, manages it, but they also have an interesting model for supervisor, vision hierarchy, So you have worker work. Worker actors, and then supervis, supervisor actors, actors that monitor them, and and will. Intervene in problem areas. That's kind of interesting architecture. What? Whatsapp also. Whatsapp you know you used earlong very successfully more recently, so the other this? Bottle called supervision, supervision trees. , lets see. So? So? So? thing. So? thing. is actors was very ahead of its time. , so? So? So? Said elsewhere, You know. this was you know now. Now nowadays we're we're very used to having you know massively distributed system. Where? appliance in our house has a little computer in it that's talking to something else, but but this was this was developed in like the early 70 S where you know there were not that this was, was was. Where? Not this was not the case. Computers were big. There were only a few of them, ? And? And? interesting how forward looking it was. i think it was. It was actually. Somewhat to too, . Ahead of its time, anyway, this, this was a volume came out in 2001, of. For Carolyn Talcott, who was one of the people who furthered theory, that had some interesting examples and. Some some pointers to. Later, later languages that use actors, there was a system called. Rio wolf. . . mentioned some earlonne. So salsa? So this is an interesting paper I found called for 43 years of actors. So, if anyone is interested in? This was for 2016 fairly recent, , For what? what the you know actual state of actor languages are, this was , This was , a very very? Survey. Survey. don't. And? And? i don't. I don't think that that well. I don't think those those details are very interesting for protocols I'll get to. I'll get to. Im. sort of im going to focus when I'm talking about here on, more on the conceptual issues which? , just. , more more interesting to me. . . then you? Hut Hewitt died few years ago, but he was, He was developing his theory, this is a late paper he wrote, let. See? See? See? the actor model with the classical turing models of computation. And he, he maintained that actors were actually more powerful than the tory model, because they, they could support nondeterminism and a few other things. , so? So as I said. To me, the the. The. The. of the of actor languages are not not that interesting. It's more i'm I'm more focused on, sort of. The kind of the use as a paradigm and , , a metaphor and the fact that there are sort of an anthropomorphic? View of computation and protocolized, view of computation, , and. And he, he would also. Also, you know, sort of it parallel to the more formal work. Was was? Was looking into that, there's a fairly well known paper called the The Scientific Community metaphor, which? Draws parallels between the way. Scientific communities work. And the way computational systems work, and the? William Kornfeld developed an Ether language which was modeled out of a house, , how science works, and the So the the. Language has a formalisms for making. Making you know proposals and. And confirm? Or, or just disconfirming certain hypoth hypotheses. , certain qualities, qualities of science are trying to translate into computational architecture like? Mo. Mo. Mo. literature, you publish something you dont well, actually, its not quite true. you dont normally retract things from the scientific lip material, that's that's more common these days, , it's it's parallel. It's plural. There's no central authority so to. Try to import some of these qualities of human systems into the computation into the computational architecture. . . fair. Welled paper was called for pewitt. it was called Offices are open systems. Group. were just trying to again, and. Group. Look at offices, offices and organizations as actor-like systems, so they're they're concurrent, they're asynchronous. They're decentralized again this. See? like sort. See? Common sense these days, but it was. More and more more innovative and or not novel at the time? And? And? people, F. Following this idea were trying to come up with coordination coordination theories for how do you model offices or how do you build? Coordination technology for for organizations there was a. Thomas Malone at Mit did a lot of work in translating some of their ideas into practical manage. Techniques. Techniques. dejong made a system called Yubic. Like the name, which was , which was also along these lines, trying to? Build. Build. build software for coordinating organizations that that use these ideas. . . . gears a little bit here. So? So? and? So? and? kind. Kind of strain of thought that were kind of branched off from this, this actor world was called Agoric system. This. This. This. might be a little bit more familiar with this because I think this had some influence on the the crypto world. , this was. From Eric Drexler and Mark Miller. Who? Who? Who? they were sort of free market libertarian types and did. Some interesting work on. On market. Oriented computing or how, how? Deas. Deas. markets could could influence computing architectures or vice versa, . And I'm not, I'm not much of a. A free market guy, but I thought I thought these papers were very interesting and eye opening, Especially this one, so this is, this is one. , I recommend people Re. Again. Again. not, I'm not sure how how interesting it is these days when i first encountered it, which was you know, a decade or two ago, it was very eye opening. it's the institutions, is abstraction, boundaries, and just pointing out. . . institutions like. Like. Like. Like. store. Fron. Fron. Fron. Fron. Fron. Human abstraction boundary between you know, consumers and the. Who are on the outside of this battery and the producers who are on the on on the inside? . . . was this was? Kind of weaving together these ideas from social economic theory and and computation, and in a way that was very novel to me at the time, I. I have a feeling that people here may have been already internalized that that notion? that's pretty commonplace these days, but it was very eye-opening to me and I always. Appreciate. Appreciate. that, give me a brand new insight, ? As they say. And? And? ideas seem very relevant to to protocol protocol thinking. A highlight of this phrase this passage, abstraction, boundaries are the key to flexibly combining complementary activities. They're talking in the context of entrepreneur and. Entrepreneurship, how people create new institutions to mesh with existing ones, I think. I think that this one, this one is is worth taking a look at. . . so? Right that this is this. This kind of insight that that technical systems and social systems kind of parallel, each other kind of. Commonplace, now there's the the famous. Law kind of formalizes that inside and there's the the example from when Jeff Bezos, you know, decreed how how? Amaz. Amaz. services would work through apis. That's kind of another. Example of of. Doing this. Anyway, I, I think. This is a paper worth worth diving into and it's it's got some reference to austri. Economic theory, which is again, not something I'm that familiar with, but i think I think it's very. Very eye opening from a protocol standpoint. . . . any questions or comments at this point, im, kind of. I apologize that this is kind of all over the place. All right, ill go on. So? So? kind of? Another. Another. of thought is that? That I find connected to this is Bruno Letour's actor network theory? ? Not just because of the name. But it? But is another? Branch of philosophy that I think is protocol relevant? I haven't haven't seen a lot of treatment of it in the the protocol community, so I kind of want to introduce that by By the way, . I'm not going to dive into these, but in in these notes i've got a lot of kind of cloud generated expansions of of. You know. You thoughts, oh. So if we want to see what? What's the relationship between lator and network thinking, I'm I'm I have my own ideas, but but actually the The? Yeah, Ai is a very good at kind of drawing out those things I've I've copied some of them here. So? and? So? And by the? Hewitt was was. Aware of the tourist theory at one point, I asked him if there was a connection between them and he said where he's implementing lator? Which? May or ma? May not have been been true, but there was there was that connection. . . Not about to. Dive into into the tour. People arent familiar with them, but yeah, yeah, has. Of social processes as involving net networks of of actors or actors, as he calls them, which can include people and institutions, and in , inanimate objects and general forces, . Their, ? And then theres some certain structural similarities with how how how act. Theory works. . . one. Difference is that act. Theory is very. And form? And lat. Lator. Lator. Lator. Lator. not very into formal models, and kind of , kind of not exa. Opposed to them, but not emphasizing them. But I'm sort of thinking along those lines myself these days, which is why I sometimes question? What? What? Proto? Ball. Ball. Ball. is or or, or? . . . . think. Letour is interesting because he. Has kind of a lets Id say a bottom up view of how social structures emerge from from the? Play of different of different interests. Proto. Proto. kind. Proto. kind. to capture and formalize that. But again it's It's sort of not more more a question, not not so question of formalism or formalism is kind of the emergent emergent property, rather than the underlying driver of these things. Anyway, so? Anyway, little Tor. Another. another figure, I think that's worth worth taking of note in the protocol World and I haven't seen a lot of treatment of him. that's something we might dive into another time so. Another. So, ? Getting to that end of this here, so the? The. The. interest, so so the act, the actor formalism. Dont think. Not. directly that. Not. That, , that useful for protocol modeling it's too. It's too abstract, too mathematical, but the basic, the the basic underlying metaphor of of. Objects or actors that that? Communicate with each other in some. Structured way that. That. That. very. That. very. well matched to protocol. Protocol thinking, and that's so my. My. My. My. suggestion for our protocol violin should be done is by using some of the existing tools of of agent based modeling, which is a kind of another, . Adjacent adja. Parad. Parad. act. . . the agent based modeling is which you know I know we've we've seen some of that, we saw it in some of the axle. Papers that? We looked at. , it's just. You could have have worlds of. Simple objects. That. A base simple rules and exchange simple messages with , with other ones that this is a. , this is an example from Net Logo Which? , is I? , guess I the the leading platforms for doing agent based modeling work and education, it's it's. it's it is, it is package. Education tool, So so it's very accessible to or reasonably accessible to to non-mathematicians non-technical people. . . i was going to show off an actual example here. let's if this works. So, yeah, this is. this is a, this is a model of slime slime, mold, aggregation. So? So? So? So? agents here, the little red dots the the. Is their chemical they're secreting? . . . . they. Simple rules which are defined in this. This. This. like language? Fa. Fa. fairly accessible, I think, and. Theres been. Theres a community of people building up very, , well, cred shot other ones and then there, there's also a kind of a visual, a visual block version of the of the language, so you could you could express behavioral rules as these simple simple constructs, so ? , you know, my suggestion for protocol modeling is have some environment where you can do do things like this, just build simple act. You know, actor based procedural models of protocols, protocols, and let them go, oh, that, should that, should that remind there was. I think here. Lets see. Yeah, there was actually. Yeah, i know we. Talk about traffic. See if this one works are people, are people seeing this?
Venkatesh Rao: Yep, we can see your screen.
Patrick Nast: Yes.
Mike Travers: Okay. Okay. heres a little traffic, Heres a little traffic light example look weve been talking about. . . not particularly complicated here, but you can change various parameters, You can see what happens with the speed limit. Goes. Goes. Goes. Goes. and again fair. Simple, fairly, simply create. That's like. you know, let's like. Like 20 lines of this logo language so? Any? Any? Any? thats my, my concrete proposal, is that? If. If. want to talk about protocol modeling? We. We. don't don't don't don't worry too much about the complicated mathematical formula, let's build some kind of procedural procedural system that lets lets people interact with with systems in in a concrete way, I think that would be an interesting direction to go anyway, that's That. That. about all I have. So, yeah? Yeah, any? Questions, comments.
Venkatesh Rao: Since that was a lot of information, and you also shared a bunch of links. Maybe we. We can do something like five. For people to kind of review the links and. Notes, and that we can do a round of comments, maybe, and also might. Might be. , page you were sharing looks like a updated version of the one you shared in the channel and it looks like that's running on your local machine, is that right?
Mike Travers: Oh, . Yeah, I think I, I. Think if you refresh that page, it might be up to date.
Venkatesh Rao: Okay, all right, so why dont we take 5 min and then do a round of comments.
Mike Travers: Sounds good.
Venkatesh Rao: Lets get started, since theres a whole bunch of links. I think the best course for each of us is to like find a couple of the links that seem worth reading further at leisure, and then we can circle back to it. But yeah, let's do a round of comments and I can. Kick us off and we'll pass up, each person pass to the next person above them on the list, skipping mic. Since he did the presentation. All right, Yeah, so thanks, Mike. That was . Illuminating, sort of like Historical Review and I. Felt that being a little too apologetic for this being an. An obsolete frame. Because it strikes me that often what happens with these very complicated idea spaces that evolve over many decades is history goes down like a particular path dependent branch. . . know delivers some ideas and. Conceptual advances there, but the branches that are not as deeply explored often they have. Differentiating elements that get forgotten and then 50 years later somebody goes back and looks at that and says hey. This alt branch actually had this one good idea, we should revive and try again, so. My, my bias in Re. History of. The. The. of a set of ideas is to be very. Open to like callbacks to really long ago and here. I think what you said got me wondering it. Feels like the mainstream history was basically. Allen, K, One, like this was happening at Mit and on the other coast, but yeah, small talk, object oriented programming, that paradigm basically became dominant, and then that led on to functional, and so forth, and even this agent pays just. Was? Was? it was pro. Even when I was in grad school in the late 90 S and early 2000, S, but it always felt like the research side branch, like through the. 90, S. there was you know, agent based programming was a thing and people were making comparisons to object-oriented and saying that it will replace object oriented programming, but that never happened. What came after object oriented was functional programming, so it feel. Like the Alt. Branch and. If I try to put my finger on the essential distinction between the mainstream and the old branch, I think. This actor and then the agent. Approaches kind of emphasized intentionality of the basic building blocks, like objects, are fundamentally reactive things right, so it kind of sits there like a lump of stuff and theres methods you can invoke on it and it has some encapsulated data, but obje. Don't want to do anything. Unless provoked into doing so, and same thing with functions, Theyre fundamentally reactive constructs, whereas actors and agents feel like active constructs, and even if it's a very simple procedural intention, they have an intention like the. Swarming Boyd's model that you showed. There, each bird has the intention of like making. Yeah. Yeah. towards the center of gravity of the local cluster, if I remember correctly, so, even though very simple. Intentionality is there, and I think thats kind of been lost on the main branch of history and. Not clear to me how you would put that back in a useful way, and whether that's a useful thing to do because I think you're right, Like if you try to apply this to protocols, i think one of the reasons this resonates a lot with protocols is. The Ent. Both there are elements that look like objects or functions that you know, push up button. Something happens. we, we talked a lot about that in the process, calculus track, and I think Giovanni is not here today but the cyberphysical systems modeling approaches that he uses. The. The. temporal logic, stuff that also has that reactive character of like put a coin in and the vending machine gives you something. But protocols, I think have that as one major side of the. Infrastructure, but the other is, in fact, something more. I hate to use the word agentic, since it's so abused now, but there's an agentic dimension to how protocols function and how they should be modeled, so I think that's worth thinking about, and yeah, i think the Bruno Latour angle is also interesting? I personally have gone down that bunny trail a bit, but it's not been super explored here. In the protocols world, we did talk about it a couple of years ago, alongside, like you know, timothy Morton, Timothy Martin has this thing called object oriented ontology, and he and Latur are both part of, I think, the Sts. Science and technology studies community, So there is a parallel humanities tradition here and? I kind of like understand. And I'm sympathetic to their hostility and skepticism towards formalism. I think I. I go back and forth, but currently, I'm on a formalist bias mode, and I think of it as yes, if you're dealing with sociology and anthropology, you should be wary of them, but when you're trying to actually design and architect engineered systems. it's actually good to like, have a formalist bias because you do need formalist. Scaffolding to build anything, even re. Remotely complex or ambitious like you can't hack and improvise your way beyond like lab, prototypes, beyond that, you need some formalism, but Ok, so those are my comments I will pass up to Robbie.
Patrick Nast: Yeah, thanks, Mike. This was a very interesting presentation. I. I get that you kind of throw cle. Little bit about , you know, keeping it high level at the beginning, but I do kind of in kind. Kind of similar to what Quentin was talking about? I kind of want to maybe ask a couple questions that? Drill slightly into like you know, actor model, like qua formalism, slightly . . . in thinking about the actor model versus something like object oriented programming, right I, I sort of the place where I struggle with this a little bit is that like? it's certainly true to me that the kind of high-level aspects that are you know? That you talked about with respect to actor like, sort of modeling concurrency, asynchrony, decentralization, Those are like, clearly like, obviously, very like useful things that we need to and incorporate conceptually into modeling protocols. I don't think that anyone would suggest otherwise, , but I'm also left with some questions about like why actor model, why the actor model you know as a formalism, or or you know, as a conceptual model is distinctive or uniquely insightful with respect to these properties, your understanding that, like as Quentin was talking about like it sort of more vaguely deeps in asynchrony than concurrency, And and maybe you know a lot of the process models kind of do the reverse, ? , and you know? For. For. For. For. we get out of, like, say, thinking more deeply about the actor model specifically that you don't get out of object oriented programming, necessarily right, like you mentioned some of the The. Commonalities right, and I think that even like the Net logo, agent based models are kind of not formally using a specific actor model, but are you know? like, kind of, basically coded in , in an object oriented programming type. You know sort of paradigm, and Im wondering Like does just. You know. You up, object based modeling kind of subsume the protocolish, insights, or is? Is there something you know unique to you know, thinking clearly about the actor model in in. In the absence of something like object-oriented programming, that's you know, really useful and important.
Venkatesh Rao: Passing up to dao.
Patrick Nast: That was. That actually a question for Mike. But?
Venkatesh Rao: Oh, okay.
Mike Travers: Oh. Oh. what? What was the question? What? whats the particular advantage of looking at actor that, the actor model.
Patrick Nast: Yeah, as Opp. Other things that would express actor, like properties, like say other, you know, , , other other concurrent computing paradigms, or like, specifically, object oriented programming right, because I also agree that agent based modeling, or like you know, bottom up style models, , you know, are a fruitful way for looking at the emergent properties that come out of protocolized systems, but like you, know, kind of as you showed in your examples like that require thinking specifically about the actor model, or is object orientation sort of enough to get the kind of protocol-ish insights is my question.
Mike Travers: Yeah, I mean, I.
Patrick Nast: My over. Question here is is at what? at what level of concreteness do we need to care about the actor model specifically.
Mike Travers: Yeah, I. I dont think I dont think you do? I dont think you do honestly, I mean that. So the actor model, ? . . think I think its useful to people who are studying you know computation and concurrency at a very you. Bas. Bas. level, I don't. I don't think it's very. I don't think it's that relevant to doing so? Bas. Protocol modeling, and kind of the large scale, sort of social social system sense. , so that's That's kind of why I was apologetic, I'm not Im not sure. i'm not sure how relevant it really is in. In, in its particular for formality, so, but it's but, but it's more, I think it's more more relevant in the realm of of how the how our general ideas about computational system have have evolved in the last few decades, . , I. , also would say like so, so you know. Actors and objected programming are not opposed to each other, I think of them, as as. Different versions of the same thing you know, actor. Are focusing on? Formalism and concurrency, the Ob. Object people are are. Are more focused on practical concerns, but they. Really different versions of the same idea, as opposed to but as opposed to like objects versus functional programming?
Patrick Nast: Sure, sure.
Mike Travers: There, there you actually have. Have have? Have of. Have of. conflict in how systems should be organized because, because, one, one of the fundamental. Both actors and objects is that state state should be distributed and. And every every Obj. Manage it. Manage own state, whereas in functional programming, is quite the opposite, ? , . , deal with this in my, my kind of. Real real world job all the time. , so, but but actor? And objects are really. You know theyre. They're sort of you know, pushing same consent you know in the. The conceptual metaphor sense theyre, theyre, just versions of the same thing. Does that answer your?
Patrick Nast: Yeah, yeah, fair, fair enough. thank you. Ill go ahead and pass it up to Dal.
Venkatesh Rao: Oh, think Dallas said that hes unable to speak right now, so passing on.
Patrick Nast: Okay, okay?
Ananth: Ill keep it short. thanks, Mike. that was a nice sort of little presentation and I didnt read the overvi. Actor languages paper. And it introduced me to this whole another way of thinking. And its one interesting that the a I laboratory of Mit published that, so thanks for that. , one thing I was noodling on was , . , like. Maybe, , you know. When you trying to, if youre trying to model like a basket of protocols, and if you can model protocols themselves as actors, some of the sort of simpler formalisms, these the? , , B. Bring about can be applied. And you can maybe perhaps distinguish the actual protocols themselves from that basket of protocols model where you can include like process, calculate. Formalism for the actual protocols themselves, so this sort of ? . . of protocols, working together is kind of where I was. Was trying to apply what I was seeing here? But But to think about, but thank you. this is very interesting.
Venkatesh Rao: Yeah, we have a few minutes. maybe Mike, you can kind of like. Share. Share. final reflections on the discussion.
Mike Travers: Sure, well. Sure, were all really good good comments? thank you and thank you for putting up with my. Kind of chaotic presentation style, . Yeah, im im . I gues. I. I. i did not dive into too much of the details of the actor formalism, and , but you may be, maybe that is worth doing. I mean I. I guess, my, my, . . . . . So, yeah, so so? . . . i. . . i. issue. Issues of of synchronization that I think I think Clinton drove up brought. Yeah, that know thats. that's where. The det. The actor formalism come come into play, and and there are in in the literature. Of. Actors, there are various ways to represent you know things like synchronization primitives. , . You know, but you know, but but buffering controls and dealing with you dealing with deadlocks, deadlocks, and such so that that might be worth worth going into at. At a later session. . . see? What? What? . Yeah, and Im still you know. Yeah, . Im still. I guess the question of what what the right formalism for understanding protocols is still. Still still an open question and? , I'm I'm still. I guess. Im still you know. A. A. and this is just because, my my back. My background is I built. Systems to do the sort of kind of behavioral modeling of like the Net logo style modeling using. You know using visual languages for. For your children, in novice programmers, so that's you know? That's something i did you know 2030 years ago, , and? Like to introduce that into the protocol community, I think it would be technical people able, some of the technical people, able, technical people able to build these. These kind of behavioral rules so. That's that's something I'm dreaming about. maybe I'll actually do do something about it, I don't know. , and yeah? Yeah, and? Yeah, and? And and again, I guess my. All. All. the stuff. Going going deep into the history of computation, I hope I hope i hope thats interesting, interesting to people. Im. Im. you know I knew knew some of these people, I knew new new Carl Hewitt pretty pretty well, actually, and. I'm
Venkatesh Rao: .
Mike Travers: I'm just. I'm just. Interested in preserving some some of this, and seeing how how these ideas, kind of you know, mutated to this very different technological and social world we we live. Now. Now. Now. Now. yeah, and and yeah, the the whole, the whole protocol movement seems to. Echo and reflect, and repeat and elaborate on some of some of these, these old I. I'm I'm. I'm that fascinating? So anyway, but thank you for listening to me, Randall about it.
Venkatesh Rao: . right, so? Thank you, Mike. That was illuminating and I think weve identified a couple of trailheads for potential future sessions. So this is the last session of 2025 and well, meet again on January 9, I believe. And hope? All have a good break and one of the things I plan to do over the break is just review all our sessions so far and maybe write like an end. End of the year review, I'll share the draft document in the channel and those of you who have things to add. Please feel free to jump in and. Add to the document and then we'll We'll probably make that a post on the protocolized magazine, so that'll be a nice little record of our first year of discussions and january 9, I believe we have a kick. Session scheduled, so Patrick and I will. Run that session in some way, we'll discuss offline and come up with a format, but we'll try and figure out like a roadmap for 2026 so? Yeah, those are the two items on the agenda year En. Review that you can all contribute to asynchronously, and then January 9, we decide what to do next year. Happy holi.
Mike Travers: Hi, everyone.