Mark Burgess (CF Engine, Promise Theory, Semantic Spacetime)
- 00:03:33 How Mark Burgess started to develop CF Engine.
- 00:11:22 What is the genesis of 'Promise theory'? What problems does it solve?
- 00:30:51 How Mark moved between academic disciplines over the years?
- 00:34:51 What is 'Smart Spacetime'?
- 00:44:29 Does the universe have a direction? What creates that direction?
- 00:55:26 Are emotions necessary for an artificial intelligence?
- 01:07:07 How can a machine memory mimic the human brain?
- 01:16:57 Can 'Promise Theory' be used to describe 'Intelligent Design'?
- 01:32:03 What are 'low hanging fruits' of scientific discovery in the next 20-30 years?
You may watch this episode on Youtube - #64 Mark Burgess (CF Engine, Promise Theory, Semantic Spacetime).
A disulfide metabolite is produced subsequently and is less active than the free thiol. Even foods high in iron, such as fortified cereals and organ meats, contain only a few milligrams of iron ( table 1). A control group of 62 patients received preservative-free saline solution as contrast agent diluent while a test group of 62 patients received saline containing 0 apotheke-rezeptfreie.com. Cellulose compounds, colloidal silicon dioxide, starch, stearic acid, and talc.
Direct stimulation of beta-1-receptors by ephedrine and norepinephrine increase cardiac chronotropy and inotropy. Check with your doctor right away if you have upper stomach pain, pale stools, dark urine, loss of appetite, nausea, unusual tiredness or weakness, or yellow eyes or skin. Available data do not suggest an increased risk of death with use of LABA in patients with COPD Beste online apotheek. Also excluded were pregnant or breast-feeding women and subjects with malignancies requiring systemic chemotherapy.
Mark Burgess is the creator of CF Engine. He is widely known for his work on 'Promise theory' and 'Semantic Spacetime'. He has published several books incl. Thinking in Promises: Designing Systems for Cooperation as well as Smart Spacetime: How information challenges our ideas about space, time, and process.
Find more episodes from the Judgment Call Podcast:
Welcome to the Judgment Call Podcast, a podcast where I bring together some of the most curious minds on the planet. Risk takers, adventurers, travelers, investors, entrepreneurs and simply mindbogglers. To find all episodes of this show, simply go to Spotify, iTunes or YouTube or go to our website judgmentcallpodcast.com. If you like this show, please consider leaving a review on iTunes or subscribe to us on YouTube. This episode of the Judgment Call Podcast is sponsored by Mighty Travels Premium. Full disclosure, this is my business. We do at Mighty Travels Premium is to find the airfare deals that you really want. Thousands of subscribers have saved up to 95% in the airfare. Those include $150 round trip tickets to Hawaii for many cities in the US or $600 life let tickets in business class from the US to Asia or $100 business class life let tickets from Africa round trip all the way to Asia. In case you didn't know, about half the world is open for business again and accepts travelers. Most of those countries are in South America, Africa and Eastern Europe. To try out Mighty Travels Premium, go to mightytravels.com slash MTP or if that's too many letters for you, simply go to MTP, the number four and the letter U dot com to sign up for your 30 day free trial. Mark, thanks a lot for coming on the podcast. I really appreciate that. Thanks for taking the time. So when I go to your website, I find this incredibly interesting. You actually have four different introductions of yourself. So one is short, one is long, one is pompous as you call it and one is more historical. Which one do you like the most? I like the shorter the better. I'm not one for talking about myself too much but there's some context in the longer ones which is sometimes helpful because it's hard to put me in a box in a particular category of things because I've wandered all over the place in my studies and in my work and so people come from knowing me from different angles and so I try to present a little bit of that history. Yeah, I think it's incredible how you say this, how understated you are and how humble you are. A lot of your co workers and people that know about physics describe you as the Richard Feynman of our generation which sounds like incredible on already. That's far too much of an honor but I'm humbly thankful and if I would hope that has something to do with the fact that I try to communicate and share what I learn in popular writings as well as in more technical work because I think it's important to share the insights that we have at a more popular level. Not everybody thanks you for that but I think it's the responsibility of science to communicate to a wider audience. A lot of people coming from a computer IT background they know you're from one of the creators, the creator correct me what's the correct term there of CF engine which is described as way to control your computer's immune system, correct? That's true. It was when I was a postdoc at the university here in Oslo many years ago, 30 years ago as many people do in the natural sciences I got involved in managing the computers, the networks and all of the networking was coming up at the time the UNIX based systems and you know I'm a person who likes to get my hands into the system and really understand everything I need to take everything apart and put it back together and make sure I understand how it works and I did that for a few months and then after a few months it started to be less interesting should we say to to spend as much time as needed to be spent maintaining systems because not everybody realizes it but the computers don't simply manage themselves they they need continuous assistance maintenance cleaning up you know garbage collection updates upgrades all of this kind of stuff and so to cut a long story short I decided it would be much more fun to try to automate that to write a computer program a smart computer program if you like a kind of artificial life kind of program to to allow the computer to become a kind of a living thing like a biological entity and manage itself as organisms to to make sure its vital functions are going well you know the heart rate is good it's not overstressed all the garbage is being collected all the blood is being filtered etc and any harmful programs or or viruses or whatever might come into or even users for that matter because humans are often the big problem all of those things are being dealt with in a real in real time because life of course is is a real time process it's not set it up set you on your on your career path and boot you out the door and everything's fine it's a continuous process of of learning adapting cleansing maintaining the system so cfng was a program that I developed over many years I wrote the initial version in just a few months to really try to turn every individual computer into its own self sustaining organism entity that would look after itself so that I didn't have to do that and then over time I realized that there was a total lack of this kind of software in the industry and I shared it as open source CERN took it up first you know the particle accelerator labs took it up first and then gradually it spread around the world and became very popular and then many years later I even started a company around this but that's the the long and the short of it yeah I think this is a complexity that few in the industry want to really look into right it's kind of a liability thing and there's just this automation complexity that people underestimate there's so many different parameters and things that can go wrong on this level right so it's usually left to human intelligence system administrator as you say to do all these tasks because I notice from my own experience if you say have the wrong version of the wrong library installed the whole system stops working and that's it's incredibly frustrating so nobody wants to really be responsible for this and it's being pushed around in IT circles from the package management pushes it to a lower category so well you have to manually upgrade and then you're like you well well it takes hours and you're like we shouldn't that be something the package management should do for instance and that's that's really disappointing so a lot of people try not to go into these levels of complexities how how big is that project now how many people use it how many downloads where they're on github do you have any of those statistics I have no idea any longer what the statistics are I think it peaked around the mid 2000s and then towards the end of I think 2008 nine time frame several competing bits of software came up some sort of using a similar approach and others going back to earlier approaches as this infrastructure IT changed significantly you know that time CFM was designed in the old workstations and servers were standalone things there were thousands tens of thousands of them at a time and that's one or two people were responsible for managing those thousands or tens of thousands but then something crazy happened you know the internet took off was ecomness suddenly every every student and their dog wanted to get a computer science learned to make a website and there was you know suddenly the market flooded with new talent in in IT of course that meant that a lot of people who didn't have knowledge of systems came into using computers so these these systems were designed for people with knowledge to express their their needs and want um sort of went a little bit over the heads of the newcomers and at the same time they had skills in programming so they wanted to to make their own things they wanted to design their own systems and started to create new systems of their own to to do this and with this you know the rise of cloud computing suddenly everybody could get hold of computers and develop software in a very easy kind of fashion so you know in the mid 2000s you could go to literally anywhere that had computers in the world turn over a rock and underneath it you would find CFM running on that computer and to some extent it's still true today you know that the giants of the industry um who shall not be named many of them still have CFM running in the enormous data centers so the footprint is there but the people actively using it sort of in a strategic way are probably far fewer today because there are many alternative ways of dealing with the managing of systems in particular cloud computing has kind of changed the way that we interact with computers and it's become much more manual again because there are far more humans to do those jobs and you know humans love to get their fingers in the pie and and mess around with it so uh there's been sort of a backlash against automation in some areas where now developers want to really have control again and push all the buttons themselves yeah the the other thing which maybe isn't as much as an accidental discovery um or an accidental venture that you got into like cf engine is promise theory right this is a big theme that you've been developing over the years maybe you can explain it a little bit to us so we have a chance to understand it why it is something that people really need to know about why it is so important um it's uh uh yeah great question but i promise there is something that i didn't intend to to create it just kind of i sort of stumbled across it in an attempt to understand the monster i created in cf engine and when i i i made the software it was very intuitive i just kind of had some intuition about how this kind of artificial organism like thing should work based on my background in physics you know my my background is in theoretical physics and i tend to see the world in terms of equilibria things in balance forces fields of influence and and so on but uh when i shifted from physics to computer science in the mid 90s i spent maybe 10 years realizing that the way that we understand systems in physics doesn't fully apply to the way that we understand systems in computer science it's for a number of reasons uh which we can get into i'm sure a bit later on but uh one of the distinctions for instance is that in in physics we try to quantify everything and we want we're looking at averages over time and over space we look for variations and trends and things like this computers are not really smooth like that they're very you know sudden jumps it's very they're a bit binary you know the ones and the zeros but things tend to be on or off here or not there in this location or in that location they're very discrete things and that smooth variation the way that we describe that in physics doesn't apply so i spent 10 years unlearning all of those methods that i was hoping to apply to understand computers and eventually kind of studied how to understand some of those quantitative things in a different way but when i'd finished all of that and actually written a book about it i realized that there was a huge topic that was missing from computing and that was what it is we want technology to do because of course you know nature doesn't have a purpose and society doesn't have a purpose it's just doing its stuff it's doing its thing and different things emerge from that and and may become sort of popular or not popular so there are shifting shifting focuses on things in in these traditional studies like physics and sociology and economics and so on but in computer science things are much more deterministic much more determined we want this to be running here we want to solve that particular problem we have this goal this problem we need to solve and here's the outcome here is the task to be done so we have specific desires or intentions that that we're applying this tooling this technology too and and that sort of desire element is totally missing from physics you know we don't physics doesn't desire anything so how do you how do you describe that how do you encode that it's not something that was is part of physics in in the same kind of way so i needed to come up with a way to to describe that intention that we have for systems what it's supposed to be doing and then the extent to which the thing that we've made the monster we've graded is aligned with that purpose or not you know is it total total chaos or is it very very precisely tracking our intentions or what is it i'm promise theory was kind of um after struggling for a couple of years trying to figure out how the heck to to describe this venturing you know into game theory and graph theory and a bunch of other things like and logic you know all different kinds of things i i really i really looked into i realized that there was no story in computer science that was really suited to that problem in just the right way a lot of things were sort of getting there but nothing was quite right so i ended up coming up with this notion of a promise which is a kind of an idealized version of the promise that we have in in day to day um which is an expression of intention intention you know i promise that this computer will uh compute this job by the end of the day or i promise that it will be running this program and it will be available for downloading this or that or you know whatever the specific intention is we need to be able to encode those intentions into i'm sorry bless you you know we need to be able to encode those intentions into the system so that we know if we're on the right path and if we're not we need to correct course correct yeah so there's this notion of drift you know systems go along they tend to drift off the path that we would like them to be drifting along and also the they tend to drift from their state of health their kind of average state of repair in which they are doing you know they have all the things necessary to complete that task so there are two ways the system can drift can drift from in the intention we have for it and it can also drift sort of dynamically because it's a bit feeling poorly or unwell or it's full of garbage or inundated with other stuff and we need to be able to course correct machines uh in both those ways so promise theory was was a way i came up with to describe that later of course you know as time went by we realized that uh probably from my because i'm a physicist and i tend to make things as general as i possibly can i constructed it as a model of agents which could be things people machines programs anything um and so it applied equally to humans as it did to machines or any kind of bit of technology it could apply to a book you know what's what does the book promise to tell you by the end of this the final page so we can apply this to the notion of promises to any kind of entity human machine or otherwise and then promise theory describes the interactions between these things you know kind of physicsy way uh how collaboration ensues once you have these basic promises which are a bit like forces or charges in physics that allow you to understand how these things can come together and work together to form systems that are larger than the sum of the parts in some sense and this sort of came about at just the right time because as systems were scaling on the internet people needed to solve exactly this problem and realized that these these issues were missing from you know the classical ways of describing software so um promise theory began to get a bit of attention still not a lot of people were interested in this but it was a lot of people to get a bit of attention slowly at first mainly in the practical uh folks you know the engineering people in in it not so much in theoretical computer science uh academia as you know is very slow to change and and a bit indignant to change if it's not one of their own ideas so uh so that's been slower but but gradually this notion of a promises found more and more applications both in um IT study of money uh economic models um and even lately and more socioeconomic things and uh studies of leadership and how companies organize their you know agile development and how companies organize lean resources etc etc etc so it's really a very general story about processes and i like this term processes because it appeals to my you know the my inner physicists uh if we can i if we can reduce everything to a kind of an abstract process whether it's executed by a human a machine a book a collaboration of animals whatever the jigsaw puzzle of how we put all these things together is is the thing that promise theory tries to describe i think this is amazing and i think looking back i feel like why isn't that the cornerstone of economic theory especially that that's kind of where i see it applicable initially because maybe that's the one i understand the best i the first thing that comes to mind when you when you told me about promise theory was either real like this part of the blockchain that specifically sets out to to have an additional contract so it's not just the token itself but there is a smart contract assigned and this smart contract is smart means it's just machines can read and understand what it is it's not necessarily smart we could say it's the unsmart part of a contract but it is a covenant that you form and and it gives you access to resources whatever those resources are and i always thought isn't that a neat way to to kind of describe or tokenize the the way how the real economy works right how real capitalism works it's this amazing supercomputer of price information and then we shift resources around based on this price information and i feel like there isn't a lot maybe there isn't i'm just too dumb to read it but there isn't a ton of research that really goes into the supercomputer it has been running on in our minds for hundreds of thousands or at least a hundred thousand years maybe with a i mean we don't need real money for it we just need like some kind of exchange exchange medium is that what promise theory and economics has been used for and people have made new discoveries or how does it apply to economics from your point of view right now when you've seen that maybe other people have picked up it's a it's a humbling experience to try to apply any any new idea i i recently realized that uh it's it's pretty daunting to imagine anyone could contribute to any field of knowledge today because so much has already been done there are so many people out there working on these things anything you try to imagine has almost certainly been imagined by somebody already somehow um in terms of promise theory it's early days especially in economics it's been far more widely used in it um let me just mention a couple of things that are characteristic of promise theory one is that it's a model of agents that are working somehow autonomously by autonomously i mean independently an agent can only promise something on behalf of itself so i i can't promise something on behalf of you because that would be imposing on you and you may or may not be willing to comply with my my wishes so i can jump things people you know can generally only promise about themselves and their own capabilities and that fits very neatly with this idea of agent based systems the blockchain the idea of independent actors collaborating entirely voluntarily in an economic or socio economic situation and to some extent this was the model that bitcoin and the blockchain folks tried to capture with this model of cryptologies there are all kinds of technological implementation details that i don't want to get into but they kind of spoil that to some extent but the the idea that essentially all promises originate from self and then need to be understood by non self you know the outside world um is kind of the essence of economics or socio economic collaboration voluntary corporation and i think of it as the default state of all all systems you know in physics we have this notion of locality that everything that happens in a in a system is kind of in in localized in a particular location and doesn't stray too far from that so it's sort of a change that happens here doesn't have a sudden effect very far away from it typically it's not entirely true but it starts from that principle and in a similar way when i make a decision about myself it doesn't have ramifications for the other side of the world without some in between process that propagates that from one side to the other and so this idea of things being very localized is is sort of at the root of it all now the blockchain tries to capture some of that by making individuals independent trying to remove the notion of a bank which is something like a relay hub in a network banks act as kind of the routers and switches of the economy when you send money out they will accept your money and route it to the to the next location and so on so the the banks sort of present the world of money as a kind of network in which money is a kind of promise you know i if you look at old currencies like the pound in the uk it still contains the text i promise to pay the bearer on demand the sum of one pound which used to mean that you could take this piece of paper to the bank of england and get the amount in gold equivalent to that doesn't really mean that anymore but it's still somehow the promise of of of value that can be exchanged for something else assuming that everyone else is along with the that that game but um uh it's curious that because of sort of his for historical reasons if you like which perhaps go all the way back to moses and the ten commandments we have this view of the world which is based more around command and control and the idea that if i do this that must happen so i push this button that must take place like i give you this money you must give me this thing in return it's entirely the wrong way to look at the world but we have this kind of bad habit of framing phrasing things in that way when i was coming up with promise to you i was fighting against that model of description of systems all the way and i think one of the things that people picked up on in the economic realm was also this fact that it describes corporation from this voluntary perspective which is more realistic rather than this force based push must happen kind of um idea that you know if if interest rates go up it must be true that this will be the effect on the economy it doesn't happen in that way of course um and there's a story around statistics and and scaled effects of all those tiny changes happening which may eventually lead to some of those things effectively happening that's a totally different kind of probabilistic story on a different kind of level and yet we need to be able to describe the one from the other microeconomics from macroeconomics and vice versa if you will and in a similar way the tiny changes on one computer to the total totality of computation computation in the cloud or the idea of a single idea to the whole of human knowledge all of these are scaling problems that we need to be able to understand and that's something that physics has a long history of being able to explain a lot of good techniques there to to draw on but the same techniques haven't really been drawn on in the same way in these other areas like computer science and economics so the folks at the Federal Reserve and a little group of researchers that sort of stumbled across promise theory and the work I did on describing money as a network they've been kind of interested in other ways that we can use this to describe a new way of describing economics based around these principles which is perhaps more realistic than these neo what's the term the neoclassical economics of of Milton Friedman and and company which are largely used by the the wider world and I think it's it's possible but no one has really shown exactly how to get that far in in that area it's still still a thing to be done but it's certainly true that the way promises work is very closely related to these cryptocurrencies that ethereum as you mentioned can be considered sets of promises encoded on this crypto ledger which you know for better or for worse are sort of immutable and will will continue to be promised forever more once you've made them which is both a useful and terrifying at the same same time so promise theory gives you a way to describe those things and understand the complexities of them and analyze them it remains there remains a huge effort to be to apply to to those things and really decide whether or not teaches you anything new or not even you come to a new discipline like that and you you're a trained physicist and I think also computer scientists you have a degree in computer science I have no degree in computer science but I've the majority of my publications are have ended up being in computer science and I have my professorship was awarded in in computer science rather than in physics at the end of the day because I'd sort of made the made the switch and and had gone into that field when you go to a new discipline and we know that is something that a lot of thinkers who think more lateral who are not boxed into specific research or professorship who have that ability to look outside the box they do complain about this compartmentalization of science and that you find it incredibly difficult to go from one discipline to another and even mathematicians have trouble really breaking into physics and physics physicists into math which seems incredible because both disciplines seem extremely related to to the outsider to seem almost the same many times and is that something that you found true as well and how do you deal with this if if it isn't as accommodating as you wish how do you how do you walk around this it's it's actually terrifying how hard it is to to flip between different subjects even when they're quite closely related sometimes because there's often a jargon associated with the with with the subject a way of formulating problems which is sort of traditional perhaps or or closely related to common practice or whatnot and there's also an enormous stubbornness in in people in one field to to not think you know to see themselves as sort of special and everyone has the kind of their own special needs and and knowledge and they don't want to be reduced to a general case of something else so there's there's often a lot of resistance from die hard academics in one area to to see what they do is being related to something else that they don't do and ironically you would think that academics might be more open minded in that way tends not not to be the case but I do find it hard myself but I've also come to it's something that I need to do on a daily basis because I I've got my finger in so many different pies that you know I'll work for a day on this problem and then suddenly I'm working on a totally unrelated thing which I find both extremely challenging and extremely rewarding at the same time because I often learn lessons about both things by by being in the frame of mind of the other while looking at this unrelated apparently unrelated problem I will see it in a new light and suddenly realize oh you know what I do know something about that it's it's just like this other thing over here convincing other people of that connection is a lot harder than seeing it for yourself I have to say so it is challenging to to to take on board and to convince other people of and this certainly applies when trying to publish results because you know journals and academics or journals tend to be tied to sort of academics who are quite powerful in their fields and they've won their fame by being successful in one area and if you try to publish some new idea which doesn't fall into that sort of classical view they may oppose that for you know political reasons or for human reasons or whatever or simply because they don't follow or understand it so it can be enormously challenging I find those cross disciplinary projects that I've worked on have been the ones that have taken longest to gain acceptance but they've also been the most rewarding ones that have had the the greatest impact perhaps and the most applicability in the long run yeah talk to Patricia Farve and she just wrote a book about Isaac Newton and I asked her about is the age of the polymath coming back and she was very skeptical she's like this maybe but there's there's nothing that suggests this right now so if anything it goes the other way right now which doesn't vote so well for for people like us I consider I'm definitely far away from the the depth of science you do but I consider myself a generalist and that is definitely well it's not a skill that's as much in demand as I would hope let's put it this way or maybe that's just me right so that there's a lot of a lot of avenues there the news thing you work on and you wrote a book about it a popular science book and it's called smart spacetime and it is about semantic spacetime both are terms that I've never heard before before we actually talked last time what does that mean is that something that that changes Einstein's theories is that a quantum quantum dynamic physics book what would should we think of when we hear the term smart spacetime so smart spacetime is the popular version of semantic spacetime that I used for my book because I thought it would be easier to understand and it turns out to have some connection to artificial intelligence and perhaps even our notions of consciousness and so on but that's let's not begin at that end this is also something that came out of promise theory and my attempts to describe ordinary processes in widely different scenarios in terms of this idea of the cooperation of independent agents and independent pieces coming together to form a whole now if you imagine you know an agent is a bit like a person it's it could be any kind of person it could be an atom it could be a person it could be a nation state it could be you know any kind of a cell in a body an animal basically the thing that characterizes all of these systems if you want to call it that at different scales and to very different in somewhat different ways but just go with this sort of general idea for a moment the thing that characterizes all of these things is that they are entities sort of localized entities that receive input from outside information from outside and they process it in some way and they may respond in some way by giving message back or generating some output making a promise a new kind of a promise delivering a service and so on you know so an atom may absorb a photon it may emit another photon or it might interact with another atom to form a molecule the molecule might interact with another molecule and form in such a way that it be for example a virus you know it may infect a cell the other one might be a vaccine which sort of neutralizes the virus or you know an antibody which neutralizes the virus neutralizes the virus so all kinds of entities can be considered from this point of view that they both do things and they have some kind of a purpose or an intention of some kind which can be aligned with and then the way that we combine all these alignable intentions forms a kind of chemistry which again allows us to combine them into new things and those new things can make new promises again they will receive new kinds of input on a new level generate new kinds of output on a new level so no matter how primitive or sophisticated that kind of model is there's something similar which is a system which receives something processes a little bit and spits it out again and what happens in between is kind of interesting of course depends on the resources inside this agent is it does it have a complex brain with a lot of memory or can it only remember a single bit you know one or a zero so the degree of sophistication matters a lot but the reason this is interesting is part physics and part computer science if you will the physics part of it is that you know we don't really describe physics in that way the way we've learned to describe physics since Newton's time Galileo and Newton told us that we have bodies entities particles if you like and they tend to move in straight lines and then they hit other things and they move off and you you follow these lines and you figure out what's going on this concept of something being inside something else or something arriving at a location and being absorbed processed and emitted isn't part of that story of physics or it wasn't at least until the quantum theory came along and then suddenly became important and so there was this kind of missing perspective in the way that we tried to deal with physics and if you look at the way people tried to deal with quantum mechanics they also tried to start with this notion of things going in straight lines at certain locations and so on and and they got into a lot of trouble with that so it took a long time to figure that stuff out on the other hand in computer science people have this very centralized notion you have a computer it has input from the outside and it generates output it's clearly centralized in a location similarly with biology you have cells that absorb and emit you have organisms that are centric that receive sense information and change behavior as a result so this centralized view of the universe almost the you know this um earth at the center of the universe the Copernican revolution that story was eradicated from physics to its detriment in some respects because we've forgotten how to understand systems in that way and go between that to this more straight line version of description and yet all of the interesting things that happen in the world are based around these more centralized configurations of information coming in process information going out so I wanted to see if we describe spacetime not in the way that Newton did as a theater in which things move around or and which Einstein took and corrected to add back some of those details which don't work out quite right when you don't take that central thing into account could you re describe spacetime by building it up from the ground level out of agents with the promises that they make more like a network so again you end up with a network description of spacetime rather than as we had for them for the money for the economic story rather than this kind of empty space with matter floating around in it what if everything is simply part of a giant network and all of these material excitations are just it's their information running around the network could you describe the world in that way when you do that it allows you a neat way of combining the semantics the meaning the interpretation of things with all of those dynamical behaviors the qualitative and the quantitative come together much more naturally in that framework than they do in either the Newtonian view which tends to eradicate semantics or the the centric centralized universe the biological point of view if you will psychological point of view even where everything is centralized and this more objective view of the world is is demoted so this constant tension between subjective and objective in the world don't go very well together except in promissory which unites them in a fairly natural way so i wanted to put those things together and try and describe spacetime and the interesting thing is that when you do that you have a story which although varies wildly in the details from the very small to the very large remains essentially correct whether you're considering an atom or a country interacting with another country or for example a human brain a conscious brain and it tells you a little bit a little bit about the necessary and sufficient conditions of what must be going on inside in order for certain promises to be kept on the outside which i think is powerful of course it sounds very easy but it's very hard to go into all those details so perhaps i make it sound easy it's not trivial by any means but it's a i think it's a story that needs to be taken seriously and so i wrote this little book trying to explain that point of view it sounds fascinating to me and i i feel like these two views on the world as you outlined them they they've vexed me to this this one part of the world where we feel even if we can't really describe it in technical terms there is a purpose there is a reason there is a creator even right something that definitely has already an inbuilt direction in life and i know you you've been quoting Goethe Sfaust on your on your homepage so he's been pondering with the same question there is we clearly feel of the we i don't know what science did in the last 200 years but we clearly feel there is this this positive direction of the universe and it goes somewhere even if the individuals don't know where it's going these are all part of this machinery they're not just a cock in the machinery and to an extent they are but they all follow a certain direction more or less and then we have this as you say this whole objective universe that is basically like cold and and it doesn't have any direction it's just there and i find and you must have thought about this if you if you go keep and if you go keep this further why is the universe there and is they a creator and well that would be the first question i would be drawn to when i would think about that theory and you must have done this probably decades ago what did you find what is your gut feeling i think we we have no way of answering the question why is the universe here to do that we would need to somehow to be outside it and we don't know how to be outside or if there is indeed anything outside of it we don't even really know where it comes from although we have a lot of stories that are consistent to some degree with things that we see within the universe you know we are trapped within this universe so our ability to get outside of it and understand that part is is is pretty limited there are still certain things that we can say about how it might have arisen but when it comes to our purpose you know this is an interesting question for me mixing together these two sides of my my work which is the physics as you say very objective and the information science which is to some degree more subjective and intention based and relates far closer to things like human human intelligence and consciousness and so on all of those those ideas what physics teaches us is that the world is very different at different scales we tend to try to look at the world from our scale our human scale of course because that's what we know and we're far better rehearsed at understanding the world from that point of view sometimes that means that we we impress upon the world things that aren't true on other scales and we make mistakes in the way we reason about it so for example we should never try to imagine that an atom behaves anything like a human being or that a galaxy behaves anything like a human being and yet the we know that the rules of interaction between things on those scales still have certain similarities there's still things to have with inputs and outputs and things that happen in between so you can you can go you can to some extent use knowledge from one scale to infer what might be going on at other scales and in physics there's this notion of dynamical similarity which expresses that point of view it's also called scaling newton would have called dynamical similarity is the idea that similar systems may have similar explanations or similar phenomena may have similar explanations in some sense and to get it right you need to adjust all of those ideas you know to make sure that you're not kidding yourself but it's basically my belief that that idea can be extended in the way that we do in physics which is purely quantitative to the qualitative level as well where semantics can also be scaled in the same way that we can scale processes and when you do that you start to see things in very interesting ways the ultimate expression of that I think sort of the grand the greatest possible outcome of that would be somehow to understand our own ability to have a conscious understanding of the world around us some people think that's impossible some people believe that there's you know consciousness is something totally different from what happens in the physical world I don't believe that I believe that we are just cogs in a machine but that we just haven't understood the extent of machinery in all of its glory properly yet we we tend to imagine that machines are more limited than they probably are there must be well my point I take it as an axiom that we are machines and yet we're nothing like you know machines that with cogs that we turn out like a car totally different animals animals things but nevertheless there are processes that take place in these things that can be described now one of the things that changes when you jump across large scales is that systems no longer are deterministic you you push the button or you turn the handle the thing the response of the system isn't 100 exactly as you hoped as you scale systems it may only be halfway or it may only be a probability of any only a chance that turning the handle actually leads to the outcome and that's not counter to the idea of of machinery lots of machines also behave in this way at scale take for example the global chain of logistics chain where we transport goods around the planet you place an order for a new car and it sets in motion a bunch of processes that result in a car being sent to you but you have no way of predicting exactly the time between you ordering the car and the car arriving at you it's an entirely nondeterministic thing even being able to being able to trace its progress around the world on shipping containers that may or may not get stuck in the sewer's canal you know that's a totally unpredictable thing because there are far too many variables to take into account and the way that we see the world is not like the Newtonian picture an enormous theater in which everything is precisely deterministic it's a subjective observation much more like Einstein's view of the universe and much more like the quantum view of the universe where the observer has a very specific point of view a very privileged position but has limited access to information and it's that incomplete information which makes the world unpredictable in our eyes from our perspective if you could imagine some kind of godlike observer as Newton did you know his his entire approach to physics was based on his belief in God that there was a creator who could see everything instantaneously we know that that's not true when you're trapped inside the universe and on different scales you have to deal with different kinds of delays and different kinds of challenges so we need to take that into account as we scale systems if you do that you find that the world behaves actually with quite predictable regularity but by predictable I don't mean we know precisely and deterministically what's going to happen I mean that we can generally write down some kind of our prediction as to what may happen and constrain that to some degree to make a kind of prediction I think we can do it in almost all cases and I find it fascinating that that is still possible regardless of the scale the place where people often get stuck and tend to object to this kind of view is the question of free will and consciousness in humans and I find this story fascinating I've thought a lot about it myself and it's I believe it to be related to this notion of smart spacetime but if you could put a boundary around sufficiently complex resources and sufficiently complex processes on the inside of some your head basically right and then all of those things can be made to work in the way that we know that consciousness works and this kind of belief that those human aspects are somehow special emotions a point of view personality that they are somehow not machine like properties I don't believe that for a second I believe that all of those things are highly natural parts of the processes if we only understand them in the right way so for example as we were as we were setting this up we were talking a bit about AI and what might be the next stage of AI some people think of artificial intelligence as being the intelligence whereby we we eliminate human emotion from the equation if you look at the history of science fiction this is fascinating notion that you know when the robots become super intelligent they will be devoid of emotion because emotion is a weakness a human failing I believe that that's exactly the wrong way around that it's those emotions that that lead to things being more important in one individual than in another individual that amplify and settle the question of should I or shouldn't I will I or won't I it's those potentials that arise in individuals in different ways because of the different subjective processing of information that leads to precisely those things the uniqueness of the individual and all of those things that we tend to associate with humanity so any story about artificial intelligence or the processing of knowledge that excludes the concept of emotion I think will fail miserably to recreate anything like what we imagine human intelligence to be I'm fully with you when I think of these parts that we can't explain in the human brain yet like emotions where they come from how they're being inherited and there's lots of them consciousness what morality ethics right I feel like when we talk about artificial intelligence those are all survival mechanisms that have made us a more successful predator a more successful animal so to speak right so that's why we we rose beyond what else is out there and we actually have a successful story to look back to the last 50,000 100,000 200,000 years whatever we want to count and I don't think that an artificial intelligence whatever the way it looks will be able to live without any of those systems yesterday might leave them behind and go to higher systems relatively quickly so it might not be 50,000 years might only be 50 years but they will go through the same problem of what are our values what are our ethics what do we prioritize how random are we as you say for that's what emotions do how how do we derive an art from and is that is a huge problem and each machine will have the same problem in a high intelligence or not and our algorithms that we evolved which seem to be the best on the planet we seem we have we seem to have that impression right now I think there's the good reason we'll develop in machines too but I think what what comes next is the problem is that we have things like GPT three who were designed for something completely different and it wasn't like a big deal wasn't like such a great endeavor was a few million dollars but it wasn't like you know Manhattan project but it suddenly could do poetry it could do html code it could do python code and even the developers were really surprised shocked maybe what it could do now it's still very elementary and it has a different approach than we would assume and it doesn't have all any of these advanced things we thought we just thought of but what people feel like there is this tomorrow you know someone in China comes up with the with an artificial general intelligence or super intelligence a human like intelligence it's it's difficult to find those words and yes they have emotions and yes they have consciousness and they will act like us so that's all good and I think we all agree on this but only I don't know 10 years later there is there's something out there that is like on a scale of several billion smarter and better developed that has all these things we've developed but it's like so far away from from our what we can see as a as a most smart human on the planet right now that we barely have any words and the the the worry is a little bit that this scales up in a in a very small time frame so we can't see it progressing because it's so quick in learning because that's what we attribute to machines or machine learning that might not be true and I hope you you can tell me better but then within a short time frame maybe within our lives and people think about singularity as a popular science word for for for describing that that in say 40 years machine intelligence is so far out that they have all this would we know but they have so much more that we can't even understand them anymore we don't know who that is they go to the stars we can't even send them letters you know there are already today plenty of systems that we can't understand so I think that's not the that's not the benchmark for creating machines that we we can't understand you know I I don't believe in this notion of the singularity for a simple reason that it it is a story again based on the prejudice of human scale scaling it it doesn't take into account the scaling of the world as we understand it it assumes that if we create a giant computer system filled with all the information we can feed into it you know we we feed all the Wikipedia into it and every bit of human knowledge that we can we can find that it will somehow behave like us but but think think about what makes a human like us we are the way we are because we're surrounded by people like us we went to school with people like us we we learned slowly over time running around a playground climbing trees and doing things that that humans do on a very specific range of scales computers have no access to that world they don't have the sensory apparatus for running around climbing trees they download information in an entirely different way from bookish sources right you can't learn to climb a tree by reading a book you do it by imitating a physical process and mimicking certain behavioral emotions through actuators that behave in a particular way creating a specific kind of information stream that we are well adapted to it would be first of all it would be nonsense to try to recreate all of those things in in IT no one would do it even if you could do it you wouldn't be able to get the same information to train it without you know literally creating a robot the same size as human being and sending it to school in the same way that a human being goes to school to learn to even care about us and our world but our world has changed so much I feel like we've cloudified ourselves so much who reads a book anymore nobody I mean nobody on the 15 ever reads a book right yes but but think about the the generation that's growing up right now who still climbs a tree it doesn't exist anymore these kids are literally in like a machine in a in a display in front of their head and they don't move for 12 hours a day that's it well now you you put your finger on something very interesting which is perhaps not so much that we should worry that artificial intelligence may exceed our best human beings that but rather that we might be turning our kids into robots by surrounding turning them into cyborgs you know by sticking their heads into these things too much of the time and and getting everything at the push of a button where you know instead of needing to use our manual dexterity to create something to use our minds we simply treat everything as an ATM where we you know I want a pot of noodles I press the button it arrives in my door you know like the the Star Trek food processor wait you don't have one of those everything we want at the push of a button is the is the quickest way to retard our intelligence and go backwards because if we no longer need to be creative if we no longer need to connect together dots causal events and we're not exercising those things that our brains evolved to solve my belief is that the the essence of intelligence not so much intelligence but well consciousness if you will is essentially a thing that our brain does to tell stories if you think of a brain as a kind of computer it's not a computer like the ones we have but in you know it's got a bunch of memory and that memory isn't moving it's you know some bits and bytes stood stored in some cell connection somehow but in order to recall those things and create a dynamical picture like for example when we're dreaming or when we're looking around us and seeing things changing in real time to create a real time picture from static memory you need a process that generates stories tells stories a narrative process and that's the piece that we're missing today we know how to take information the outside pattern it and store it in a static way in order to compare to patterns detect fingerprints faces etc so we know how to do the things that eyes do and the parts of the brain that that decode the patterns that senses decode and turn that into static imprints we don't yet know how to read reconstruct dynamical worlds from those memories in the way that we do uh when we're sleeping for example and when we're asleep the main difference is you know brains are very active the main differences were cut off from our senses to a large degree and so the things that we're imagining can only be coming from the memories inside us and yet what bizarre things that we we dream and it seems so real when we're there in the moment you don't question the reality of it when you're dreaming only later when you you you compare it to the world that we were more used to anchored in reality do we start to question or that probably wasn't real it's a different kind of thing but that that illusion of reality that we experience in dreams i think is the is the crux to understanding the our conscious understanding of the world the ability to tell stories and and that again is it comes back to this promise semantics based on my dear that we need to combine dynamics with semantics patterns meanings with changes and that's how we create stories that's what we mean by stories right the a stream of semantics is basically a story so that's i think the the great challenge for us to to approach some kind of a generally intelligent whatever that means organism and i think it'll take a lot longer than 40 years to get there because i think we just don't understand scaling in it there are two parties looking at this question i i read a lot of books by both of them so on the it side you got all the ai folks believing that the neural networks have some magical mystery that will just sort of make it happen i don't believe that for a second because they don't understand scaling and then you got physicists on the other hand all these usual suspects who fall straight into the trap of physics nv wanting to talk about entropy and information and and quantum computing and all of these things i don't believe that's got anything to do with it either because that's about an entirely different level entirely different story where semantics have no place so unless you unify these two pictures the semantics and the dynamics and something call it smart space time or semantic space time if you will basically agent models and networked structures network structures the ability to constrain and allow freedoms in a constrained way around localized processes individuality different viewpoints all of those things have to come together then only and only then will we start to understand what we think of as um as consciousness you know um there a lot of people would say well when we talk about meaning it's literally just and i know this word doesn't fit there a compression and an index and algorithm so instead of saving the whole image which is maybe say in one gigabyte a big picture or a couple megabytes we just say well that's a picture of a dock and it's a happy dock and 15 other attributes so we just compressed it from a huge data set to a really small one so that seems to be something that's going on in our mind right so we we only we save the shapes and we shape we save certain semantics meanings attributes and then we have an imperfect indexing algorithm it seems to be something that that is relatively obvious to be solved with artificial intelligence because it's good at tagging stuff it's good at at compressing it to two very small number of attributes and once you have in a model that's the fascinating thing you can apply it once you feel it works for what you want to do and for your data set you can apply it anywhere you want it's really cheap it's really fast processing it only takes my milliseconds but building the model takes a long time and that's kind of what when we describe ourselves as humanity we built the model that takes forever but once you figure out what we want to do and how we classify things it's relatively quick I think this is what artificial intelligence researchers right now on one hand they know what they do is just pre or statistical models and it's really boring and it does there's no consciousness into it but they feel like once we make that step that we can basically classify everything or compress so much then the next step comes in and we have like this next level of AI or maybe a slightly different technique and we feel like well we even with relatively limited amount of space we can make sense in the sense of we can make a proper probabilistic description of how this what this could be and we can predict the next next step so when a cat you know what the cat probably does it's an easy model to do and that seems to be something that even children have they know they can't distinguish spiders from cats for instance and you don't even have to run them through a big model they come with the model already that's it's in their brain when they when they are born maybe maybe they learn it over time but it doesn't seem to be such a big stretch and then to to assume that we can describe the world and build an intelligence that can go through the world kind of like we do with an imperfect model it's never going to be perfect we add some error correction like we do for children right don't touch the oven it's hot that figure they figured this out pretty soon not to do this again attention model I feel all the the switches are in place to get to something that resembles consciousness even if it doesn't have consciousness but then again we look at people the scale of consciousness in humans is pretty enormous like we find people who are pretty obliterated what's going on in the world if no clue they don't want to know about it and then we find other people who philosophers who spent day and night just you know sharpening the knife of consciousness so there is a huge scale but if you say we we just want to build an AI that for now works like a I don't know eight year old I don't think they're that far away from it yeah you I think you touched on an interesting point an important point which is that a key aspect of processing information is the ability to compress it into tokens or symbols symbolic and there's this artificial distinction in AI groups between symbolic AI and non symbolic AI or neural networks typically machine learning if you like I don't see those things as being distinct I see the machine learning tools as being ways of compressing information into certain representations which some of which are smarter than others in the sense that they can adapt or embody variations in a more interferometric way it's kind of like an interferometry and the the mistake I think is in believing that you can reduce all of the cats and catness to a single symbol or word you know and it's like in a book you read the word cat your mind immediately thinks of a million cats and all the crazy things that cats do but there is no sense in which having in gone in the one direction allows you to go in the reverse direction we don't know how to do that yet and I think some of the most advanced brain researchers of today are starting to talk about the brain not as being something which is mainly fed by inputs but actually something which generates the world from within and rather anchors those imaginings those processes on the interior anchors those things to sensory uh interferometric sensory inputs combinations of sensory inputs and then tags those things in some smart way because we don't know how it works yet and the ability to generate stories and recreate those worlds is again part of the problem that I think is missing but um it's a really important thing that a cat it's not a it's not a symbol it's a it's a process the cat process it probably more like a living thing in our minds than it is a static memory and I don't believe we've begun to tackle that problem yet you know my hobby is ever since I was a small kid the first thing I ever wanted to do in my life was music I was always fascinated with the orchestra and and the way that an orchestra comes together from all of the individual instruments and each individual player is playing with great skill and great individuality but the sum of all of those processes creates music on another scale which is the symphony itself which has a lot of interwoven parts themes interweaving with counterpoints and all of these complex musical things which I love to do as well I I compose music as a hobby as well but I see very much physics information AI all of these topics are very much like music in that sense that it's the coming together and interfering of processes rather than bits of static data in the data model being looked up by a machine like retrieval process so it's unlike that ordering that car from the other side of the world through th