on [not] making tea

By chance, when looking for information on the film that usually appears on top of tea brews (!), I came upon this highly ranked blog entry of a security expert explaining how not to make tea. Which did not seem completely right in my tea-oholic eyes..! Not that the following rambling is of any relevance whatsoever!

On the agreement side, it is indeed hard to get decent tea in most places, the primary reason being a lack of understanding that very hot water is needed. The worst being these cafés where they bring you a cup of (definitely not hot) water with a tea bag on the side! I used to travel with my own kettle to avoid this issue, but I am striving to carry as little stuff as possible and hence gave up on that habit. Instead, I often take a thermos bottle that contains an infuser: all that is needed is hot water!

On the disagreement side, the obvious resolution of most complaints about poor quality tea, “herbal teas” that are not tea, tea bags in general, &tc., is to carry your own loose tea. It is light and keeps well and cannot disappoint. And can be brewed several times, especially oolongs. The section about milk is beyond discussion as tea with milk is another beverage altogether. I certainly enjoy drinking duh-wali-chai  in India and am even making some at home from time to time, but otherwise I stopped putting milk in my tea during the first COVID lockdown. (Which also considerably simplified my tea consumption when travelling: all that is needed is hot water!) The main issue is however in using boiling water. Which is almost never recommended for brewing tea! Especially green and Darjeeling teas. Instead of using water above 90⁰, one should stay below 90⁰… Especially when running several brews. Not only this keeps the bitterness under control but it avoids loosing oxygen and CO² contained in the water.

As an aside, this film/sheen is the result of “an interfacial reaction of polyphenols and other components in the tea that bond with ions in the water”.

conditioning on insufficient statistics in Bayesian regression

“…the prior distribution, the loss function, and the likelihood or sampling density (…) a healthy skepticism encourages us to question each of them”

A paper by John Lewis, Steven MacEachern, and Yoonkyung Lee has recently appeared in Bayesian Analysis. Starting with the great motivation of a misspecified model requiring the use of a (thus necessarily) insufficient statistic and moving to their central concern of simulating the posterior based on that statistic.

Model misspecification remains understudied from a B perspective and this paper is thus most welcome in addressing the issue. However, when reading through, one of my criticisms is in defining misspecification as equivalent to outliers in the sample. An outlier model is an easy case of misspecification, in the end, since the original model remains meaningful. (Why should there be “good” versus “bad” data) Furthermore, adding a non-parametric component for the unspecified part of the data would sound like a “more Bayesian” alternative. Unrelated, I also idly wondered at whether or not normalising flows could be used in this instance..

The problem in selecting a T (Darjeeling of course!) is not really discussed there, while each choice of a statistic T leads to a different signification to what misspecified means and suggests a comparison with Bayesian empirical likelihood.

“Acceptance rates of this [ABC] algorithm can be intolerably low”

Erm, this is not really the issue with ABC, is it?! Especially when the tolerance is induced by the simulations themselves.

When I reached the MCMC (Gibbs?) part of the paper, I first wondered at its relevance for the mispecification issues before realising it had become the focus of the paper. Now, simulating the observations conditional on a value of the summary statistic T is a true challenge. I remember for instance George Casella mentioning it in association with a Student’s t sample in the 1990’s and Kerrie and I having an unsuccessful attempt at it in the same period. Persi Diaconis has written several papers on the problem and I am thus surprised at the dearth of references here, like the rather recent Byrne and Girolami (2013), Florens and Simoni (2015), or Bornn et al. (2019). In the present case, the  linear model assumed as the true model has the exceptional feature that it leads to a feasible transform of an unconstrained simulation into a simulation with fixed statistics, with no measure theoretic worries if not free from considerable efforts to establish the operation is truly valid… And, while simulating (θ,y) makes perfect sense in an insufficient setting, the cost is then precisely the same as when running a vanilla ABC. Which brings us to the natural comparison with ABC. While taking ε=0 may sound as optimal for being “exact”, it is not from an ABC perspective since the convergence rate of the (summary) statistic should be roughly the one of the tolerance (Fearnhead and Liu, Frazier et al., 2018).

“[The Borel Paradox] shows that the concept of a conditional probability with regard to an isolated given hypothesis whose probability equals 0 is inadmissible.” A. Колмого́ров (1933)

As a side note for measure-theoretic purists, the derivation of the conditional of y given T(y)=T⁰ is arbitrary since the event has probability zero (ie, the conditioning set is of measure zero). See the Borel-Kolmogorov paradox. The computations in the paper are undoubtedly correct, but this is only one arbitrary choice of a transform (or conditioning σ-algebra).

Insane craving for food

Within a couple of weeks, I read two related US stories about ordering food from an insanely far destination, like hand-made frozen pizza from Napoli, Italia, or like one startup called Goldbelly ships frozen food made by some restaurants nationwide. (With a motto of Whatever [food] they dream of, wherever they are.) While I am not consistent in consuming non-local food and drinks, like my mass orderings of Italian wines and Darjeeling teas, and while I’d love to get a new taste of Toukoul’s Ethiopian dishes, a creamy sepia risotto from Da Franz, an okonomiyaki from any street stall in Osaka, and many many other dishes from all over the World, it sounds to me rather debatable to have a special single meal prepared on the other side of the World and delivered immediately to one’s table… Furthermore, one of the perks of dining at fine restaurants is exactly to dine at fine restaurants, not in one’s own room, and having starred chefs’ dishes ending up in reheated frozen plastic containers is certainly killing a major share of the experience.

smoked tea

The shop of the tea dealer Nathmulls in Darjeeling burned down last week. In possibly suspicious circumstances… While they lost at least 2,000 kg of their tea stock, and most sadly someone died in the fire, Nathmulls can still deliver orders, including great 2019 first flush Darjeeling teas.

Castleton second flush 2018

After receiving an email from the tea dealers Nathmulls in the town of Darjeeling for an offer of the latest summer harvest of F.T.G.F.O.P. Castleton tea, I made a join order of two kilos for my te-addicted colleagues in the maths department in Dauphine, delivery that I found in my office this morning. Truly terrific tea, at about a fifth of the price asked by the major tea chain in Paris!