05/28: Shaw

[shevajia]

Comment below with questions or thoughts about the reading for this week's workshop.

Please make your comments by Wednesday 11:59 PM, and upvote at least five of your peers' comments on Thursday prior to the workshop. You need to use 'thumbs-up' for your reactions to count towards 'top comments,' but you can use other emojis on top of the thumbs up.

[bhargavvader]

Hello Lynette - I still remember meeting you in that brewery in Santa Fe last year and talking about complex systems and sociology. It's so cool to read your work a year later, and to have you present here at the Computational Social Science workshop! I enjoyed reading both the papers and am looking forward to hear you chat about it with us.

Your exploration of the way value emerges in complex socio-economic systems through agent based modelling was the first time I've seen something like that - to be honest I don't often see Agent Based Modelling outside of SFI related folks (and some computational archaeology and anthropology literature), probably because of how it may be difficult to verify the results against a dataset: the predictive paradigm we see in most CSS papers these days. My first question is in general about such models: how do you see them working in comparison to the more statistical, predictive nature of ML? Sure, ABM is, in the end, an iterative process with our bayesian priors updating, which while similar in some ways to ML, has not quite caught on the way ML has (or maybe I am reading different kinds of literature). While your work convinced me, do you think that not having a "performance" against curated datasets a drawback of such models? Or are you more concerned about questions of where it gains stability and the ways even a single agent can affect this?

My other question is about how we can incorporate Content Analysis techniques with ABMs. I enjoyed your Grounded Approach (Nelson 2017), and I think that STMs can be very useful in revealing structure in unstructured text. Do you think such information can be incorporated into models, where agents make decisions based on the text generated by other agents? So we will be simulating what folks were talking about on those sub reddits and forums, while seeing if the value of BitCoin matches the one simulated by the model. I can think of a couple of ways this can be engineered, but I'm just not as fluent in the ABM literature to say for sure. In general I'd love to hear more about your experiences of using ABMs in sociological analysis and the pros/cons.

[nwrim]

Thanks for an interesting read, Dr. Shaw! As a person who has small to no exposure to cryptocurrency, I found your draft highly amusing. I agree that Bitcoin could be a natural experiment on how "value" originates - although I am not sure if "value" of currency based on the Austrian principles (those who are supported by nations) and "value" of cryptocurrency are in the same dimension (or the dopamine and "wanting" psychologist/neuroscientists talk about), but it is interesting research nonetheless.

I think your research quite convincingly shows the transition from theories of money (why is this worthy?) to practical necessities of adoption (how can we make Bitcoin more useable outside our core communities?). And I agree that this was a "critical part of Bitcoin's success" (p. 24) - but my question is that how did this happen? In other words, what led to the formation of "communities' shared belief that Bitcoin had value" (p.35)? Was there a symbolic event where people saw that Bitcoin indeed could be turned into real-world goods (like the pizza buying example you gave, however, I think that event is being joked and referred as "the most expensive pizza in history" in some communities I saw) or did the Bitcoin had too many users that it reached a point where it is "too big to fail"? I know that this could be out of the scope of the research, but I wanted to hear your insight about the issue.

Also, I think @bhargavvader's questions are golden - I would love to hear more about agent based modeling and ABMs (although I do not have much experience with both).

[bakerwho]

Hello and thanks for joining us this week.

Your work in economic sociology is fascinating to me in how it adopts two scales that are not commonly seen used together. You explore both, the ways a conceptual framework can be computationally modeled, as well as the ways computational models can unveil rich structure and insight from massive data.

I'm particularly interested in the latter (wrt your draft "The Inevitable Sociality of Money"). Do you have any tips to share with us on algorithmically guided explorations of large datasets? It seems to me like this might be an important paradigm for future social science research - to use methods like topic models or network analysis to flag features of interest in massive datasets.

On that note, a more pop-science-of-science question - what are your favorite topic modeling algorithms and why?

[liu431]

Thank you for presenting. In your draft "The Inevitable Sociality of Money", some of the latent topics seem to be correlated with each other. It is hard for readers to examine the similarities and differences of a large number of clusters. On idea is using topicCorr to produce a correlation graph of topics. Do you have any other suggestions to effectively visualize the results from STM?

[policyglot]

Dr Shaw, Thank you for this interesting juxtaposition of sociology with economics and Bayesian ABMs. You touched on how inaccurate estimation of value emerges in the situation of bubbles. But for financial markets more generally (pre-Bitcoin), how would explain the emergence of derivatives trading? After all, it seems like the core creation of value in a Marxian framework has long been lost under layers of financial abstraction.

[hesongrun]

Dear Professor Shaw, Thank you for the presentation! It is really thought-provoking to examine under what conditions the social consensus will converge to the 'intrinsic value' of an object, which has been the center arguments of the market efficiency. It is also interesting to see how our society constructs value out of Bitcoin. However, I am a little bit confused about the definition of 'intrinsic value'. How do you define it? What do you mean by the non-socially originating part of the valuation? Why do you use the 'bootstrap' in the title for the construction of economic value? Thanks!

[SoyBison]

Dr Shaw, Thanks for sharing your paper! I always love to see A: Agent based simulations, and B: scientific approaches to philosophical issues, and the papers you shared combine these things wonderfully. My question is a tad abstract, but bear with me. At the root of many social science questions there is some kernel of insolubility, they often leverage some assumption that can be challenged, and has been challenged, on philosophical grounds. Your work, for example, deals heavily with value theory, but many other social science questions rely on claims about ontology, ethics, and so on. The converse to this is also true, especially recently; some philosophers have leveraged social-scientific data to support their claims about the nature of humanity, and especially ethics. The work of John Doris for example.

I think in your case, this is a particularly interesting situation. In your papers you address abstract aesthetic concerns with concrete modelling and theory. Generally the hard-soft spectrum in academia is not bridged like this, with social-scientific theories that rely on concrete modelling generally being used to address concrete problems, and vice versa. In a sense this makes your work extremely cross-disciplinary, but in a different sense than how we usually think of the term. So my question is this: What is your approach, as a scientist, to this scenario? How do you think this approach can be abstracted to the social sciences at large? In other words, how would you describe your utopian Science?

[vinsonyz]

Hi professor Shaw, Thank you for your presentation! Your paper builds a formal understanding of how real value is capable of precipitating out of social action. In your perspective, how would your work pave the way for future theoretical and empirical studies in such field?

[PAHADRIANUS]

Thank you for sharing your brilliant analysis on the formation of cryptocurrency's value. As you pointed out, the emergence of the likes of bitcoin is partly a renewal of metallism and Austrian and right-wing economic ideals. You have elaborated greatly how cryptocurrencies, despite their vast differences with traditional money and inconsistencies among their own creators, are embedded with value. Still, I find them more similar to the valuable tokens such as gold, stocks, or stable foreign currency in relative to an inflated domestic currency, rather than money in daily circulation. Does this mean that there is something different in their value formation in comparison to that of the ordinary money? You mentioned how state force and institutions play roles in valuating money, but could there be other additional reasons that ordinary money, now not just in the form of banknotes and coins, is still dominant?

[goldengua]

Thank you so much for sharing your research on how formal bayesian theory would inform our understanding of social values. I was in particular curious about how this bayesian could potentially be linked to our social belief in conversation exchanges. For example, in the referential task where the speaker need to refer to an object to listener. Speaker not only need to choose some features that match the target object, but also need to select distinctive features based on speaker's understanding of how listener would interpret his utterance. I was wondering to what extent social values could evolve through interaction similar to what happens in conversation exchange?

[lulululugagaga]

Thank you for your presentation Dr.Shaw. This is a pretty up-to-date study. I like your idea on how bitcoins are interesting as research objects to study the origins and formation of 'value'. I think it would be also cool to study the adoption/spread of such electronic currency worldwide.

[skanthan95]

Thank you for sharing your research! I would love to hear more about your methodological rationale. As I understand it, structural topic models are distinct from other topic model types like LDA and NMF. I was wondering why you chose this specific type of topic model over other options!

[di-Tong]

Thanks for sharing these interesting projects! I wonder if you could explain more on how the focus transition in the Bitcoin’s online communities happen. What are the conditions that prompt such transition and Bitcoin's success afterwards? Do you see similar processes for other virtual products?

[hihowme]

Thanks a lot for your presentation! This is a truly interesting topic about applying the computational methods to dig some really important questions about the value. I have a question that besides currency(bitcoin), the comments and reviews nowadays are having more important impacts now, with many people are hired to make comments and leave good reviews online. do you have any idea in how to define the "intrinsic value" of those text and word of mouth?

[sanittawan]

Thank you very much for sharing your research with us, Dr. Shaw. Sorry if my question is very rudimentary as I am not familiar with your area of research. While reading your paper on the Bayesian learning model which studies the social construction of value, I keep on wondering what kind of situation or context the simulation in the paper can be situated in (aside from cryptocurrency). For example, does it apply to how traders react in the stock market during the IPOs? Could you please give an example of the context where your research may apply?

[bjcliang-uchi]

Dr. Shaw, this is a very, very interesting paper on using Agent-Based Modeling to construct the value-making process in a random society. My question is about the agent networks in your model. The paper says that learning agents then learn by beginning with "agents randomly select without replacement a sample of n other agents in the system" and update their values. I am not convinced by this core methodology because in reality, people's communication largely depends on existing networks defined by power, kinship, knowledge, culture, and race. In fact, it is exactly these determinants which are considered by sociologists as the determinants of social values. If the model does not assume the existence of these determinants at all, how would this result be considered as relevant to the reality?

[ruixili]

Thanks for your presentation and wonderful papers! I am really interested in these two papers. I guess my question for the first paper is that how do you validate your theory? And the question for the second paper is that how you related this one to the first paper? Thank you!

[RuoyunTan]

I think it is very interesting to analyze peoples' values through mathematical construction. This is something that I do not have experience in. Looking forward to tomorrow's presentation.

[Yawei-Li]

Hi, thank you for presenting in advance! As a layman of cryptocurrency, your work really opened up my mind. I also really enjoyed the theory-crafting of linking money and currency theories to the technologies part. Looking forward to see you tomorrow!

[KenChenCompEcon]

Thanks for your presentation! I am curious if the method can be applied to the prediction of future social valuation development, under special events such as Covid2019? Thanks

[heathercchen]

Thank you very much in advance for your presentation! I have a quite general question for your paper. Since we are not able to fully look into your dataset, I am wondering about the theoretical construction of the paper. In other words, how do you make sure that the data you have about Bitcoin and the corpus is valid and robust to let us infer to the social-economic basis of it?

[TianxinZheng]

Thanks for presenting your two interesting papers! Mathematical construction of value is really interesting work. I was wondering how do you choose the specific topic model to use in the bitcoin paper.

[minminfly68]

Hi, thanks for presenting this interesting topic. However, as some other students mentioned, before, could you classify more about the agent-based modelling and the intrinsic value attached to it? Thanks again!

[lyl010]

Thanks for coming! I am interested in the interpretation of values: the values is about the economic values and how to we model the formation some values like, aesthetic values? Thank you.

[ShuyanHuang]

Thanks for sharing your work with us! In your paper on Bitcoin's value, you ran a series of structural topic models and used the R function “exploreK” to identify the best justified choice of the number of topics. Did these models (with slightly different k and different initializations) produce stable results?

[SixueLiu96]

Thanks for your presentation! Bitcoin is a really hot topic. I wonder would you mind explaining more the relationship between virtual currency like Bitcoins and real currency? What will be for the future development of virtual currency?

[jsgenan]

Thank you for sharing with us two interesting works! It is moving to see a delicate scientific analysis of the philosophical valuations behind practical moves. In the example of Bitcoin forum, you emphasized the importance of "practical affirmations of worth" in the evolution of valuation. I was curious about the users' entry or exit during the era of confrontation. How much of the shift in affirmation was due to users changing their stand, how much would be creditted to old members quitting the discussion?

[fulinguo]

Thanks very much for your research. My question is related to the psychological foundations of your findings. Could you please discuss more about how the analyses are related with human behavior sciences? How this research could provide guidance on future studies in psychology? Thanks so much!

[linghui-wu]

Welcome to our workshop, Dr. Shaw! It's very exciting to read your paper about Bitcoin. My question is similar to the one proposed by @hesongrun. Could you elaborate more on the meaning of "intrinsic value", such as its source or how it is generated? I believe it would be beneficial for me to better understand your work!

[ellenhsieh]

Thank you for presenting such interesting topic! In the paper, you mentioned that people’s perception towards Bitcoin value continue to diverge. However, I am wondering how you deal with the people who do not believe Bitcoin possess any value or those who have no knowledge of Bitcoin in your analysis. Did you just ignore those observations? If so, did it cause any bias in your research?

[Yilun0221]

Thanks for the presentation! Same as Songrun @hesongrun and Linghui @linghui-wu , I did not quite understand 'intrinsic value'. I think this is quite an abstract concept in politics science or philosophy. However, I think it is very cool to do quantatative research on abstract concepts! In addition, I wonder how did you decide to use the models, like why Bayesian not Markov Model?

[sunying2018]

Thanks for your presentation! It is very interesting to understand Bitcoin from practical affirmations of worth. I am interested in the “computationally grounded” approach and the two phases framework of analysis. I have a question about the topic models that if a corpus involves many relevant topics, which is not rare to see, how could you identify the final viewpoint of this corpus?

[MegicLF]

Thanks for your presentation! I am quite interested in your paper about Bitcoin and I am curious about how you identify the internal/collective impact of previous posts about understanding Bitcoin have on the later posts. Besides, since the understanding of the value of Bitcoin is pretty vague, I wonder how you think when starting this project and how you quantify those discrete concepts.

[SiyuanPengMike]

Thanks a lot for your interesting and inspiring paper. In your paper, you demonstrated how to calculate the economic value of bitcoin, which is quite impressive. Besides all features you included in your model, I'm wondering if other natures of those digital currencies will affect their economic value. Will the power of the platform behind those currencies be a major influential factor? Thanks again for your cutting-edge topic!

[YanjieZhou]

Thanks for your presentation! Idid content analysis on large scale data before and found it really hard to gain some useful information from those unordered texts, like the concept of intrinsic values. Do you have some experience to share about the specific methods to explore those datasets in an economic or sociological way?

[harryx113]

Thank you for the presentation. I think one of the reasons why we can't put a rational price on bitcoin is that we can't factor in people's confidence on a decentralized monetary system, greed, demand from the dark web. Do you think we will ever be able to put them into an equation?

[dongchengecon]

Thank you so much for your presentation! In the research, you have adopted a two phase computationally grounded method for the analysis. I am wondering if there are other potential computational design that could also achieve the goal of balancing the benefits of large-scale statistical analyses with the richness of qualitative analysis? And you may use that framework as a robustness check for your main result.

[Panyw97]

Thank you for your presentation! It's so interesting to explore the issues related to Bitcoin. Could you elaborate on the relationship between virtual currency like Bitcoins and real currency? What will be for the future development of virtual currency?

[tonofshell]

Thanks for sharing your work! You mentioned that traditionally most research on value has been done qualitatively, with your research positioned as a compliment to it . Do you think that qualitative and quantitative research on value are fundamentally in different lanes, or will future research will allow for more integration between them?

[yutianlai]

Thanks for coming! I wonder if you could explain the meaning of "intrinsic value" further. It is a very interesting concept to me! Thanks!

[caibengbu]

Thank you for your presentation! It's so interesting to explore the issues related to Bitcoin. Could you elaborate on the relationship between virtual currency like Bitcoins and real currency? What will be for the future development of virtual currency?

[adarshmathew]

Thank you for your presentation, Prof. Shaw! Your methodological paper on Bayesian Learning Agents was a fascinating read, and reminded me why I was first enamored with ABMs and Bayesian Learning. If you ever release the code for this paper, I would love to reconstruct and tinker with it.

Two questions on model perturbations and one on model extension:

1. Constrained communications via network structure: In addition to the three scenarios you model for, I'm curious how your results would change if a network topology of communication was overlaid onto the agents in your simulation. In the current setup, value declarations at each stage are public (or semi-public within a neighborhood). Would overlaying restrictions on communicating value via a network structure significantly change your results? We could take different cases: highly centralized networks, networks with multiple cliques, etc.

2. Learning Rate as a parameter: Another key perturbation I could think of is asynchronous learning rates of agents -- how would convergence change if agents update their beliefs asynchronously, with some agents updating every time step and some others doing so every 3/5 steps?

3. Modelling Preference Falsification games: Timur Kuran's concept of Preference Falsification in his book 'Private Truths, Public Lies' is something that constantly haunts me every time I come across polls and work around stated beliefs of agents. Would it be feasible to extend your model into a two-phase game where agents have public and private beliefs which influence each other, but deviance from the Mode public belief is penalized?

[boyangqu]

Thanks for your presentation! There are a couple of other digital currencies similar to bitcoin, such as Litecoin and Libra, is the valuation rationale applicable to other bitcoins with different context?

[HaowenShang]

Thanks for your presentation! I am wondering why you choose to use Bayesian Agent-Based Modeling rather than other modeling methods to conduct the research on investigating socially constructed valuations. What’s the advantage of using this modeling method?

[YuxinNg]

Thanks for sharing your presentation. Since you are doing researches on bitcoin, which is not only a hot topic but also a sensitive topic in many ways. So I am wondering if there is any legal or ethic concern for topic like this?

[luxin-tian]

Thank you in advance for your sharing. Bitcoin is a interesting topic for me. What are the potential social consequences it may cause and how can researchers examine it in a more general perspective?

[JuneZzj]

Thanks for presenting. It is good to hear many fascinating terms such as animal spirits. In the discussion part, you conclude that this research is associated with the dynamics which emerges when the mixture of social and non-social sources are involved in the collective process. I am not sure what kind of specific real implication of this concern. Thanks.

[Anqi-Zhou]

Thanks for your presentation in advance! Could you please explain more about the generation of Bitcoin's "intrinsic value"? And besides Bayesian Agent-Based Modeling, what other kinds of modeling have you tried?

[timqzhang]

Thank you for your presentation ! It is really interesting to see the quantification and modeling of the "value". I have the similar question as others mentioned, that is the virtual currency actually have a lot of forms right now, so I wonder if the modeling could also be applied to the ones other than the bitcoin?

[cytwill]

Thank you for this presentation. I am really interested in this Bayesian Agent-Based Models, especially how does this differ from other forms of agent-models. Hope you would explain more about tomorrow, thanks~