hi everyone my name is i am an architect and a researcher by training i also teach at ucl fyi i am not a software architect i am a building architect which is not so different principally as both are trying to put ideas out there to maximize choices within the system so my work stands at the convergence of virtual and physical infrastructures and today i'll be discussing with you the application of real-time media and algorithmic personalization in urbanism a live stream economy but before i do so i'd like to give you a little disciplinary context which is going to show you in a more concrete manner why live stream economy so in the space where climate change mitigation and building design overlap there is green architecture which is an approach to building that minimizes the harmful effects of constructions on our environment nonetheless the spectrum of color should really be expanded to accommodate more than just one class of building that passively assures ecological health but a full category of functionality that actively recovers our geo conditions for instance there should be a spectrum of grain architecture that is not only designed for energy efficiency but actively functions to facilitate let's say regenerative agriculture there should be red architecture which functions for the support of forest fire resistance blue architecture for ocean iron fertilization yellow architecture for carbon capture storage ccs brown architecture for soil carbon sequestration and so on and so forth these are but some examples that i'm putting on the table for now to start the conversation but the spectrum should really be expanded to accommodate many more colors so there are three main questions that my research tries to investigate first how much human architectural infrastructural and economic capital will such an active decarbonization system demand let's take the yellow architecture as an example which is carbon capture storage ccs according to gene holy book under our 2 degree carbon budget we will have to capture and remove 10 gigatons of co2 in our atmosphere by the end of the century which is 2100. to capture one gigaton would demand about a thousand facilities to capture tangible we will need ten thousand facilities starting from this moment presently we have about twenty two of them so we are already far behind schedule this means we will need not only architecture but infrastructure on the order of the existing fossil fuel industry it took more than a century for our oil industries to get to where we are today but we don't have another century to lose we will need decarbonization strategies that can generate profit as much as fossil fuel if not more in order to realize this amount of work our architectural pedagogy is at the forefront so the second question is how many do we have to educate in order to decarbonize so that's an economy of architectural education in itself we will need a large capacity of architects who are equipped with such forms of interdisciplinary knowledge on the other hand the design of decarbonization system is an interdisciplinary problem thus we will need means to search for tools that falls between the established fields of expertise promotes cross-exchanges cross-communication and cross-stacking last but not least different countries have different leading carbon capacities due to their differing available natural and artificial resources for instance sweden has about 4.5 gigatons of co2 storage space whereas norway which is just next door has about 21 gigatons capacity potentially the two countries can negotiate on carbon trade have some of sweden's shipped to norway and have it stored under the sea if we look halfway across the earth china has about 65 to 1500 gigatons storage capacity as you can see the gap for approximation is relatively high the reason behind is that there are always unpredictable fluctuations in geological and demographic factors and you can imagine the larger the country the larger the uncertainty also all these numbers that i pulled from different research they're all projected based on different estimated scenarios so all these will have to be calculated and indexed according to variables such as mass of population how much energy his country is exporting in the support for global consumption and so on so it's not as easy as saying that you know this one country seems to be over budget on numbers and we just focus on regulating her but more like how do we have a planetary scale collaborative index on such numbers according to what variables what scenarios so see these are some important questions to ask for any initiative that wish to participate in the carbon economy and have all these data transparent and immutable so as to incentivize international exchanges to distribute both rights and responsibilities have risk allocation over large-scale collaborative contracting thus we would need means to enable notes across geographical and territorial boundaries to communicate in real time and most importantly is to prevent shadow economies to prevent situations like you know some countries exporting the co2 to less privileged ones off the book so instead of solely depending on regulation it's also helpful to think about why and how businesses would be willing to engage and do things the proper way meaning how do we incentivize international deharmonization supply chains how can they generate profit and all together how can we gauge supply and demand more accurately within our trades so this is the same on an industrial level different industries have different latent carbon capacities for instance the carbon footprint for computing can be mitigated relatively easy with solutions such as solar energy but things like airplane transportations aviation has much lower latent capacities simply because there is for now no alternative energy source but fossil fuel so there's a lot of talk about carbon trading but it's actually very challenging to sense and regulate carbon emission suppose we can have every factory to install and live stream their by spectral camera even if we do that it's still quite inefficient so perhaps it's helpful to think again from an incentive point of view about how industries can share not just the load of carbon with each other but the load of work where together they can maximize the total payoff so to maximize coalition in cooperative games for instance to share the load of work of the tourism industry with the livestream industry while both maintaining their increase in total productivity so virtual space dependency has always been controversial and may even sound offensive to some for instance it's a common volcanology that automation and competition may take over jobs of humans especially undermining jobs that are dependent on our building environment like tourism but after what we have experienced over the past year we may begin to reframe these sort of questions differently for instance reframing questions like how can we make trade-offs between tourism and live streams to asking questions like how we can make more payoffs with tourism live streams so that the livestream space may help to sustain the tourism industry and help out-of-work labors within economic crisis this question definitely goes beyond the pandemic and help us in starting to think about the demanded changes to our modes of working for the situations to come which i'm sure is going to be increasingly challenging this is where real-time media may be helpful even before the pandemic the livestream economy in china alone has already overcome hollywood as one of the largest media industries but its intrinsic properties has potential to go beyond entertainment because of its participatory authorship so livestream is a form of decentralized media that enables a feedback of information in real time in terms of architectural design the same logic can be applied to virtual desktop so how can more than one person work on the same architectural model especially with complex 3d models that take gigabytes of data how can we have preemptive scheduling to test which amongst a network of computers what is the amount of control each node should have how do we trace and track all the work input so that's participatory authorship via streaming most importantly how can we direct royalties to content creators and contributors instantly much much like how we're able to do via livestream platforms so there is an expansion of value networks livestream expands the rigidity of advertising and subscription models that are used to finance digital platforms to a processing of micro payment transactions in real time for instance in china there is a tradition of red pockets which represents the virtual of luck and digitization enables such forms of micro values to be sent back and forth between parties in real time via diverse value rounds and that generate cash flow for all kinds of entities from individuals to culprits most livestream platforms design ar or interactive elements for every transaction so such forms of virtual signaling acts as marginal utility that fosters an attention economy so this is a significant financial model that we can think about when considering decarbonization where this microservices architecture can be personalized for diverse carbon business supports of course it is no news or that we're living in a world where our everyday lives are guided by the screen time of scrolling between algorithmically recommended content and personalization can help us to tackle problems of information overload within an infinite scroll of live streams but it can do more for us than just preemptive marketing to facilitate participatory design strategies when we talk about a collective contribution of information there has to be ways in which we can ensure the interoperability of information at the point of creation to foster a richer common data environment cde but within the design industry which largely operates on information asymmetry like copyrights maintaining financial incentives for qualities in work input within additional common is of equal importance it's actually one of the greatest challenges like how do we make sure designers are willing to contribute to a common pool by getting paid accordingly so if our goal is to enable a multi-access system for decarbonization there has to be ways in which we can self-organize for instance ways to interact in real time and rich consensus on the permutation of architectural information where the value of such permutation can be realized for each individuals here i use a work of mine to illustrate what i mean which i recently published on the on the topic of coupling blockchain's shared data layer with beam's common data environment the question is is it possible to index information objects in a decentralized manner amongst a scattered chain of designers and clients much like why commercial search engines like google are doing but instead of a proprietary ranking individual should have some sort of participatory control over such ranking you know like a collaborative certain process with real-time data streams or incorporating a folding mechanism to filter out excessive noise within the information system so this helps to prevent the creation of common data environments that translates pinterest architecture into pinterest urbanism so that's really a problem we have to have to think about in terms of recommendation and personalization together real-time media and algorithmic personalization can help us to foster a livestream economy for decarbonization with a focus on the following realms first education and interdisciplinary research second share the load of work from space from space and time sharing to work sharing third is to generate profit so my approach to thinking about this wicked problem which i consider a design problem is that if there is a soft infrastructure that can aggregate not just our data but our computing power a solved infrastructure that help us to circulate units distinct from monetary values so from gdp to gdc02 and to trade in a way that the more we exchange let's say the healthier the amazon forest will be not in the sense that we just buy more starbucks coffee and donate to brazil because that only adds layer of mass on top of our accounting profile so this is some trees being printed on a paper cup can you believe that but to trade with units they are directly tied to information on the forest so digital currency is definitely something to think about where blockchain is a good way to tokenize but that's just a matter of compute before that we would need to deal with sampling so we would need a network of sensory devices infrared remote sensing low-cost multi-spectral optical cameras and so on anything you name it so that we can see the health of the forest we will also need live stream on such data so information can be distributed better yet the live stream is entertaining so it catches not just large capitals and large investments but aggregating the efforts of even you and me for instance you know as an average citizen why would you pay for access to some tree data well how can looking at some trees be entertaining that's within the expertise of the media industry one successful example is alibaba as you may know they have different kinds of digital wallets one of which is the end credit pay so that functions like a digital credit card instead of monthly payment and expensive transaction fees you can pay back your debt on a daily basis and that boils down the time intervals to cash flow and because they focus on micropayment it's just easier for non-bank account holders to acquire credits so it's more far-reaching to rural areas even indigenous villages they also have a doula ball there's no english name for it basically it's an entertainment valve from taobao what it does is that it acts like a savings bank with the extra money you have in your taobao digital wallet you probably don't want to keep transferring between a wallet and your bank account so taobao offers you to save that money with them and they will give you that 7 interest rate in return by venture capital into the entertainment industry and media industries so that's very livestream economy the ants grad credit together with your tower wallet which partner with many many many local businesses they have this tree planning scheme where if the app senses that you lowered your daily carbon footprint for instance if you use the subway with a wallet it will earn you carbon credits which you can make micropayments to plant a tree and you have this digital version of your realtree which is very kawaii to interact with you can also see all of your friends treat trees so that builds up a carbon fiber network and it's exactly a form of virtual signaling as of today alibaba and its users planted about a billion trees in china with all these live streaming and micro payment schemes also they really focus on working with technologies that are both technically and financially available for democratization i am going to briefly introduce two of our ongoing design projects that embodies the spirit of the aforementioned research first is a speculative urbanism project current that thinks about the future of broadcasting cinema and its impact on our cities so me and my collaborators experimented firsthand with a range of emerging technologies all are readily available to any individuals and proposed a production pipeline that facilitates the collective production processing broadcasting and archiving of urban data in real time so livestream includes image and metadata which can be extracted for environment and event reconstruction with machine learning we can have estimation on what's behind a foreground object so that is perfect to be coupled with photogrammetry frameworks which calculates based on vantage points on the other hand we experimented with ai image processing like autoencoder which can help to fill in missing information on texture maps based on archived data while object detection can help to estimate scene descriptions so the output volumetric data will then be plugged into personalization algorithms which will label rank and deliver recommended contents by collaborative filtering finally the output will be pulled into displays on demand which are volumetric navigation engines like vr devices and access by a network of users so this basically helps to reconstruct 3d environments based on multiple vantage points from sequences of 2d images especially with the ever mounting terabytes of image data we're producing via social media so this is a demo we produce in the most economic way possible for participatory purposes so this facilitates an attention economy where the reconstruction of certain events and environments may direct value back to the entity for instance we reconstructed polar bear tracks so a polar bear who walked through mountains of trash because he couldn't find enough food in the polls using livestream data from bear camps we tried to harness network effect so this can potentially generate financial and social credits for the protection of the species of course this stays as an artistic expression for now but it tries to illustrate how such system may look and feel like based on this we're currently working with 21e8 the magic magic number company to realize a blockchain camera system where chains of cameras can authenticate urban events amongst themselves to increase our capacities in retaining evidence for the much contested event landscapes in our 12 percent potentially we can have micro licensing anchored on parts of the entity be it human or pulse human entities like trees or the north pole for instance we can have the 3d scan of a polar bear let's call the polar bear sam we can have some micro licensed and whenever sam or part of sam generates a profit either received a red pocket from live stream or someone purchased the license for use micro values will be directed to ngos for the protection of species while all transaction data is transparent and immutable on the blockchain so that we would be able to know whether the money is indeed used for the bear so this is some architecture stacks we're proposing based on john nash's theory paper in agencies models and these are my collaborators on this project we come from very different disciplines and different countries where livestream and algorithm personalization actually got us together you can get in touch with us by the link below and follow up on our progress second project is candide city adversarial network design so this is a project that formulates a new real-time visualization toolbox for architectural and urban design the project considers the participatory authorship problem and aims at collapsing the linearity of pre and post-production also human to machine design flow currently the process of decision making in design involves repetitive negotiation and contractual matters when we change a small bit in a design it changes the entire digital model even if we use parametric tools which which helps to automate alterations we will still have to re-execute the visualization and that takes a lot of overtime and a lot of interns people think that's what architects do but no this is what architects do imagine if it's on the scale of city planning changing the building height of one small area we probably have two teams of interns working two more weeks of overtime to redo all the calculations and visualizations also translating from 2d to 3d has been a fundamental problem in architecture since the time of brunoliski so this is called the figogram problem and is the define of the boundary between spaces often used to analyze private to public sounding ratio besides functional it is also an aesthetic concern a perceptual grouping problem which is perfect to be coupled with machine vision so we try to utilize this into our technical framework which aims at formulating a computationally cheap toolbox using lightweight processes to enable a real-time collaborative modeling so there are a lot of attempts for virtual desktop and architecture but for modeling it's especially challenging let's say we want to put a building information modeling system on the blockchain with decentralized computation but such pipeline is just not very affordable outside central hubs at least for the time being it may work on the digital model with just a few megabytes but b models easily reach over a gigabytes or more so we are trying to find alternative solutions of course we're not trying to solve the problem overnight but we're taking small incremental and economical uh steps as designers starting with a figure ground drawing which can be acquired from any open source map box the differences in color indicate the differences in building heights so that naturally embeds 3d information in a 2d drawing we took a small portion of the city and utilized a rule-based procedural processing on mesh tension analysis to generate a range of new city grids each has small articulated differences based on parameters we can also do this with style transfer algorithms but for the aim of this project we do it the cheapest way possible computationally so what's the difference in the generated grids well in the fields of urbanism the original grid is called a functional sony which promotes hygiene and highways i mean it can be good for corona but occasionally it promotes the death of american cities like detroit because of the segregation in spaces the generator grids are estimations between this and a more organic arrangement so the mesh tension analysis helps to identify areas where it's too densely packed with high rises and those which are too segregated this may help to provide options with better ventilation and circulation strategies to assess decision making in urban design so this functions a bit like an architectural machine as proposed by nickrpante in the 70s that the designer or any user can become a curator on aesthetic decisions or become a system designer who formulates parameters and boundaries to the system while the machine automate the processing and recommend adjacent articulations personalized to your demand so of course that's where decentralized indexing would be useful within a database of digital assets like architectural objects and building components now comes the biggest challenge how to visualize it in near real time so we can communicate the design efficiently to a scattered chain of clients and contractors the drawings already has building height data embedded all we need is to extrude them using architectural softwares we can do this at the snap of a finger like using c brush or whatever so that is not a problem the whole digital model is labeled with colors for machine vision for that we still need some manual power but we're working to figure this out potentially with a big enough set of synthetic data we may be able to train a neural net for it so then we flew a virtual camera around the city model and we plugged the image data into open source ai algorithms in this case generative adversarial neural network thus the project name this particular one is trained on german city data we can also swap in a hong kong dataset or anywhere else on earth that we need and based on semantic labeling it outputs a video visualization with picks to picks so the one i'm showing you here is very computationally affordable it compresses what would take days of work into just hours with a fully trained neural net so there is also pix2pix hd which offers higher resolution but we're leveraging between cost availability and quality for the needs and capacity of different design works for the next step of the project we're thinking to expand it to a city-scale semantic labeling and personalized data sets to which we may begin to classify stuff like wind corridors transportations infrastructure provisions different types of zoning and so on and so forth to perform visualization for a broader range of design needs we are also testing out integrating other open source ai into the pipeline like game reader for static visuals so here are my collaborators on this project we are based around the world and livestream is one of our most used tools in co-working so you can count on us that we will keep pushing to develop our work from virtual space sharing to time sharing anyhow i normally do a little takeaway but i don't have time to summarize today but i will do it on instagram so make sure you smash that subscribe button i'm going to end with a very cheesy tagline the future is not urgent because it is already here but our action is urgent we've got to collectively reconfigure our modes of working living and even lecturing where urbanism and ways we organize directly influence how we'll mitigate our climatic crisis by transforming our econometrics from gdp to gdc02 so one small incremental step at a time and we will get there if you're interested in any of our projects or wishes to contribute to the conversation feel free to send me an email you can slide into my dm on instagram as well so you'll be able to find me at p r o v i d e s dot ism thank you so much for listening
2021-01-18