This YOUTUBE link is shareable.

Watch and comment LIVE or view the archived episode!

Subscribe (free!) to the @PodSaveChocolate YouTube channel, like this video, comment, and share this episode to help grow the #PSC community.

Watch this episode on YouTube

This LINKEDIN link is shareable.

Watch and comment LIVE or view the archived episode on LinkedIn.

Join my network on LinkedIn to receive notifications and to refer business to each other.

Watch this episode on LinkedIn

This FACEBOOK link is shareable

Watch and comment LIVE or view the archived episode on TheChocolateLife page on Facebook (for 30 days, then watch the archive on YouTube).

Follow TheChocolateLife on Facebook to receive notifications and catch up on other content.

Watch this episode on Facebook

Episode 197 Overview

In this post in the How To Be Cultured section of the paper of record (The New York Times), we learn that “We asked 43 culinary professionals to tell us their favorite brands available in the United States. Below, their top nominations under $25, from least to most votes.”

In this episode of PodSaveChocolate we’ll examine the claims and how they final list of twelve was arrived at. But first, let’s take a look at the background of the author (at the NY Times).

What I couldn’t find (in LinkedIn or anywhere) is that the author has any subject matter expertise in the area of food or chocolate, and there’s nothing in her bio that suggests she has any specific background in maths, statistics, or survey study design. EVERYONE has to start somewhere, and someone at the NYTimes greenlit this article, which involved reaching out to more at least 43 people (there might have been some who did not respond).

There’s no indication of how the contributors were selected, what they were asked (specifically), or if one of the two chocolate makers mentioned was prohibited from voting for their own bar.

The selection criteria as I do understand them:

1. Under $25.
2. Must be available in the United States.
3. At least two of the 43 contributors must have nominated it.
4. Each nomination must be “a favorite brand” of the contributor.
5. The interviews were “edited and condensed.”
6. Assumption: The contributors were not picking from a list.

Evaluation Process

The page, as it is organized, is a mix of different underlying presentation (HTML coding) formats. So, first, I needed to figure out which data had to be extracted, how to extract it, and then get the data into a form (or forms) that that made comparison and analysis possible

The Only Chocolate Bars Worth Buying

We asked 43 experts to tell us their favorite brands available in the United States. Here, their top nominations under $25.

The New York TimesMackenzie Oster

I was forced to approach extracting the data from the page in steps:

1. I copied and pasted the list of contributors into an LLM chat pane and asked the LLM to parse it. (Pro tip: tell the LLM parser to key in on the semi-colons as the delimiter.)
2. Once I had a parsed list (below), I asked the LLM a) to figure out the company affiliation of the contributor and where (city, state, country) the company was located, and b) find a URL for the company, of if independent, for the contributor (not shown). Those tasks turned out to be difficult, and I had to manually hunted down the affiliation for about 1/3 of the list using a combination of search techniques.
3. I used an LLM skill (/web-scraper skill --llm-extract) to extract the data from the twelve bar recommendations in two steps, describing exactly was I was looking for each time. (Pro tip: stepwise refinement makes things go faster and reduces the likelihood that the LLM will return erroneous results – they do make mistakes, often very stupid ones, but sometimes very subtle ones.) The first step grabbed the company name, country, number of votes, and price, generating a table. The second step grabbed the comment (if there was one) and the name of the contributor, adding it to the table.
4. Once I had the two lists complete I queried each with a series of prompts. For example, for the Contributor Analysis table, I asked the LLM to generate a count of the number of times the phrase “chocolate maker” (and “chef” NOT “pastry” or “restaurateur”) appeared in the description, and a calculate a percentage of the overall total. Note: I have to check if the LLM included the column names in the total (44 rows instead of 43 rows when calculating the percentages.

What can we learn from analyzing the data on the page?

Contributor List

Contributor Analysis

What can we learn about the “chocolate appreciation level” of the contributors from the following list?

The NYTimes’ list of “The Only Chocolate Bars Worth Buying”

Vote Distribution

In Conclusion

1. Why did I spend my time examining this post?
2. What was the quality of the data on the page?
3. What was the point of this exercise on the part of the NY Times? What does this tell us about their editorial intent and process?
4. Does this advance the appreciation and advancement of chocolate?
5. In the final analysis, is the headline clickbait? Are these twelve bars the only ones that are actually worth buying?

Future Episodes

Episode Hashtags and Social Media Links

#NYTimes #listicle
#cocoa #cacao #cacau
#chocolate #chocolat #craftchocolate
#PodSaveChoc #PSC
#LaVidaCocoa #TheChocolateLife

#PodSaveChocolate and #TheChocolateLifeLIVE Archives

To read an archived post and find the links to watch archived episodes, click on one of the bookmark cards below.

Pod Save Chocolate Calendar and Archive

News, views, and conversations on topics in cocoa and chocolate streamed live to YouTube, LinkedIn, and Facebook. #PodSaveChocolate!

The Chocolate LifeClay Gordon

#TheChocolateLifeLIVE Archive

News, views, and conversations on topics in cocoa and chocolate streamed live to YouTube, LinkedIn, and Facebook.

The Chocolate LifeClay Gordon