I am a Singaporean PhD candidate in Philosophy at UCSD.

I work mostly on philosophy of physics and philosophy of science, with particular interests in the philosophy of thermodynamics and the conceptual knot that is entropy, and the use of approximations and idealizations in physics. My CV can be found here. 

In Winter and Spring 2022 I will be in University of Illinois Chicago as a pre-doctorate fellow in the philosophy of quantum gravity, working with the Cosmology beyond Spacetime project funded by the John Templeton Foundation.

On the side, I dabble in formal epistemology, data ethics and the philosophy of data science, and the philosophy of logic and mathematics. 

I completed my undergraduate studies in Philosophy at Wolfson College, University of Cambridge, and spent some time at the Munich Center for Mathematical Philosophy before arriving at UCSD for my PhD. 

For what it's worth, my Erdös number is 5.

When I am not doing philosophy, I can be found playing Magic the Gathering at my local game store, playing video games on my console/PC, or skateboarding on the streets. 



2018 -

PhD Philosophy, 
University of California San Diego

2017 - 2018

MA Logic and Philosophy of Science, 
Munich Center for Mathematical Philosophy (Incomplete)

2014 - 2017

BA & MA (Cantab) Philosophy,
University of Cambridge



Degeneration and Entropy

(2022, in Lakatos’s Undone Work: The Practical Turn and the Division of Philosophy of Mathematics and Philosophy of Science, special issue of Kriterion: Journal of Philosophy, edited by S. Nagler, H. Pilin, and D. Sarikaya.)

Lakatos’s analysis of progress and degeneration in the Methodology of Scientific Research Programmes is well-known. Less known, however, are his thoughts on degeneration in Proofs and Refutations. I propose and motivate two new criteria for degeneration based on the discussion in Proofs and Refutations – superfluity and authoritarianism. I show how these criteria augment the account in Methodology of Scientific Research Programmes, providing a generalized Lakatosian account of progress and degeneration. I then apply this generalized account to a key transition point in the history of entropy – the transition to an information-theoretic interpretation of entropy – by assessing Jaynes’s 1957 paper on information theory and statistical mechanics.

No Time for Time from No-Time

(with Craig Callender, Philosophy of Science 88:5, 2021)

Programs in quantum gravity often claim that time emerges from fundamentally timeless physics. In the semiclassical time program time arises only after approximations are taken. Here we ask what justifies taking these approximations and show that time seems to sneak in when answering this question. This raises the worry that the approach is either unjustified or circular in deriving time from no–time.

Does Von Neumann Entropy Correspond to Thermodynamic Entropy? 
(Philosophy of Science 88:1, 145-168, 2021)

Conventional wisdom holds that the von Neumann entropy corresponds to thermodynamic entropy, but Hemmo and Shenker (2006) have recently argued against this view by attacking von Neumann's (1955) argument. I argue that Hemmo and Shenker's arguments fail due to several misunderstandings: about statistical mechanical and  thermodynamic domains of applicability, about the nature of mixed states, and about the role of approximations in physics. As a result, their arguments fail in all cases: in the single-particle case, the finite-particles case, and the infinite-particles case.

Improving LIME Robustness with Smarter Locality Sampling 
(2020, with Sean Saito, Rocco Hu and Nicholas Capel, AdvML '20: Workshop on Adversarial Learning Methods for Machine Learning and Data Mining, KDD2020, August 24, 2020, San Diego, CA.)

(Video of talk available here.) 

Explainability algorithms such as LIME have enabled machine learning systems to adopt transparency and fairness, which are important qualities in commercial use cases. However, recent work has shown that LIME's naive sampling strategy can be exploited by an adversary to conceal biased, harmful behavior. We propose to make LIME more robust by training a generative adversarial network to sample more realistic synthetic data which the explainer uses to generate explanations. Our experiments demonstrate that our proposed method demonstrates an increase in accuracy across three real-world datasets in detecting biased, adversarial behavior compared to vanilla LIME. This is achieved while maintaining comparable explanation quality, with up to 99.94% in top-1 accuracy in some cases.

An Empirical Route to Logical ‘Conventionalism’
(2017, In: Baltag A., Seligman J., Yamada T. (eds) Logic, Rationality, and Interaction. LORI 2017. Lecture Notes in Computer Science, vol 10455. Springer, Berlin, Heidelberg.)

The laws of classical logic are taken to be logical truths, which in turn are taken to hold objectively. However, we might question our faith in these truths: why are they true? One general approach, proposed by Putnam and more recently Dickson or Maddy, is to adopt empiricism about logic. On this view, logical truths are true because they are true of the world alone – this gives logical truths an air of objectivity. Putnam and Dickson both take logical truths to be true in virtue of the world’s structure, given by our best empirical theory, quantum mechanics. This assumes a determinate logical structure of the world given by quantum mechanics. Here, I argue that this assumption is false, and that the world’s logical structure, and hence the related ‘true’ logic, are underdetermined. This leads to what I call empirical conventionalism.

Is Logic Empirical? Logical 'Conventionalism' from an Empirical Standpoint
(2017, Aporia Vol. XVII, Recipient of the Aporia Essay Prize)

The laws of classical logic are taken to be logical truths, and logical truths are taken to objectively hold. However, we might question our faith in these truths: why are they true? One often avoided approach is logical conventionalism, because it makes the logical truths dependent on somewhat intersubjective linguistic conventions. Another approach, proposed by Putnam (1975) and more recently Dickson (2001) or Maddy (2007), is to adopt empiricism about logic. On this view, logical truths are true because they are true of the world alone – this gives logical truths an air of objectivity unlike logical conventionalism. Putnam and Dickson both take logical truths to be true in virtue of the world’s structure, and the structure of the world is to be understood to be given by our best empirical theory, quantum mechanics. As it turns out, the structure of quantum mechanics apparently makes true the laws of quantum logic, and falsifies (one half of) the distributive law, something which was taken to be a logical truth under classical logic. Empiricists take this to indicate that the distributive law was not a logical truth to begin with. However, this argument assumes that there is a single determinate structure of the world prescribed by quantum mechanics. In this essay, I argue that this assumption is false, and that the structure of the world is underdetermined in quantum mechanics. Likewise, the choice of ‘true’ logic, as given by the world’s structure, is also underdetermined. This leads to what I call empirical conventionalism: the world alone fails to determine our logical truths. We need something broadly intersubjective, and thus less than objective, to fix our choice of logic even under empiricism. An attempt to avoid one form of conventionalism has thus led us back to another.


Drafts and Works in Progress

Do Black Holes Evaporate? (First draft in the works, please contact me if interested!)

Since Hawking first predicted that black holes lose mass and `evaporate' via Hawking radiation, the phenomenon has become a linchpin of black hole research. However, I will argue that we lack justification for thinking that black hole evaporation occurs. The derivation of black hole evaporation requires some global notion of energy conservation, which in turn rests on assuming that the spacetime in question has certain idealized properties; the ubiquitous ones are stationarity and asymptotic flatness. By examining these idealizations and how they cannot be appropriately `deidealized' in describing actual physical systems, I argue that we lack justification for concluding that actual black holes evaporate.

On the Status of Temperature and Thermodynamics in Relativity (In Preparation)

Taking the formal analogies between black holes and classical thermodynamics seriously seems to first require that classical thermodynamics applies in relativistic regimes. Yet, by scrutinizing how classical temperature is extended into special relativity, I argue that it falls apart. I examine four consilient procedures for establishing the classical temperature: the Carnot process, the thermometer, kinetic theory, and black-body radiation. I show how their relativistic counterparts demonstrate no such consilience in defining relativistic temperature. As such, classical temperature doesn’t appear to survive a relativistic extension. I suggest two interpretations for this situation: eliminativism akin to simultaneity, or pluralism akin to rotation

Indeterminism for Nomological Bohmian Mechanics
(Under Revision)

Given the nomological approach to interpreting the wave-function in non-relativistic Bohmian mechanics, and box-standard notions of time-reversal invariance and indeterminism, I argue that Bohmian mechanics is indeterministic. On the one hand, (1) Bohmian particles have deterministic trajectories. On the other hand, (2) Bohmian mechanics is time-reversal invariant. However, given (3) the nomological approach { on which wave-functions are interpreted as (part of) the laws, I argue that (1) and (2) cannot be true at the same time. At least one of (1) - (3) must go. I consider and reject four options: giving up time-reversal invariance (for the theory, and `partially' for the wave-function alone), giving up deterministic trajectories, and revising our definition of determinism. In the end, I conclude that we should abandon the nomological approach instead.

Accuracy, Entropy, and the Free Energy Principle
(In Preparation)

I show that the relative entropy – also known as the Kullback-Leibler divergence (Kullback-Leibler 1951) – can be construed as a normative measure of epistemic accuracy in the veins of Joyce (1998). Furthermore, I discuss the Free Energy Principle – made popular in recent years by neuroscientist Karl Friston (cf. Friston 2007) and philosophers like Andy Clark (2016) – and how it employs a formally identical measure, though as a descriptive measure of perceptual accuracy in neural processing. I then propose that we naturalize – and provide naturalistic grounds for – the normative measure by appealing to its realization in our brain.