<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>taste on S Anand</title>
    <link>https://www.s-anand.net/blog/tag/taste/</link>
    <description>Recent content in taste on S Anand</description>
    <generator>Hugo -- 0.156.0</generator>
    <language>en-us</language>
    <lastBuildDate>Sun, 15 Feb 2026 14:12:28 +0800</lastBuildDate>
    <atom:link href="https://www.s-anand.net/blog/tag/taste/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>How to develop taste</title>
      <link>https://www.s-anand.net/blog/how-to-develop-taste/</link>
      <pubDate>Sun, 15 Feb 2026 14:12:28 +0800</pubDate>
      <guid>https://www.s-anand.net/blog/how-to-develop-taste/</guid>
      <description>&lt;p&gt;Developing taste &amp;amp; judgement are an essential skill in the AI era. &lt;a href=&#34;https://www.nber.org/system/files/working_papers/w24626/w24626.pdf&#34;&gt;#&lt;/a&gt; &lt;a href=&#34;https://arxiv.org/abs/2309.12338&#34;&gt;#&lt;/a&gt;&lt;/p&gt;
&lt;p&gt;&lt;img loading=&#34;lazy&#34; src=&#34;https://files.s-anand.net/images/2026-02-15-how-to-develop-taste.avif&#34;&gt;&lt;/p&gt;
&lt;p&gt;But taste is different from knowledge and takes more time.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Gaining knowledge is a software upgrade. It strengthens existing synapses. It&amp;rsquo;s fast, reversible, no new &amp;ldquo;cables&amp;rdquo; required.&lt;/li&gt;
&lt;li&gt;Taste is a hardware upgrade. It destroys inefficient pathways, grows neurons for new pathways, and wraps axons with myelin speeding up signals 100x. (London cab drivers &lt;em&gt;literally&lt;/em&gt; have a larger hippocampus.) &lt;strong&gt;Taste takes time&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;How we acquire taste depends on the environment.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;In kind environments (with clear, immediate, accurate feedback, like sports, surgery) is easier. Practice at the edge of competence. &lt;a href=&#34;https://daviddidau.substack.com/p/the-problems-of-practice&#34;&gt;#&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;In wicked environments (investing, hiring, politics, strategy) confidence can be misleading. So, &lt;strong&gt;audit prediction reasoning&lt;/strong&gt;: write your predictions &lt;em&gt;with&lt;/em&gt; reason. Months later, if you were right for the &lt;em&gt;wrong&lt;/em&gt; reasons, treat it as failure.&lt;/li&gt;
&lt;/ul&gt;
&lt;hr&gt;
&lt;p&gt;Practices that help:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Taste requires complexity &lt;em&gt;and&lt;/em&gt; understanding &lt;a href=&#34;https://tiffaning.com/files/Silvia_2013_Interested%20Experts,%20Confused%20Novices.pdf&#34;&gt;#&lt;/a&gt;. So, &lt;strong&gt;when bored, complicate; when confused, study&lt;/strong&gt; by copying, comparing, asking why, and prototyping a vocabulary. &lt;a href=&#34;https://commoncog.com/the-tacit-knowledge-series/&#34;&gt;#&lt;/a&gt; &lt;a href=&#34;https://youtu.be/k4oC7_cAvBU&#34;&gt;#&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://rationaleonline.com/docs/en/tutorials&#34;&gt;Argument Mapping&lt;/a&gt;: create granular mind-maps of arguments, find hidden assumptions, and evaluate evidence.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://www.psy.lmu.de/isls-naples/intro/all-webinars/collins/cognitive-apprenticeship.pdf&#34;&gt;Watch experts&lt;/a&gt;: watch experts at work, guess their next moves, explain your reasoning, and copy but with extra constraints.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://talks.ui-patterns.com/videos/badass-making-users-awesome-kathy-sierra&#34;&gt;Perceptual Learning&lt;/a&gt;: learning by comparing examples and prototyping a vocabulary.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://hbr.org/2008/09/how-pixar-fosters-collective-creativity&#34;&gt;Brain Trust&lt;/a&gt;: have peers critique your work &lt;em&gt;against a goal&lt;/em&gt; at early stages (to focus on core, not polish).&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://en.wikipedia.org/wiki/Mindfulness&#34;&gt;Mindfulness&lt;/a&gt;: to reduce sunk-costs and other biases/blindspots. Helps with post-mortems.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://kk.org/thetechnium/scenius-or-comm/&#34;&gt;Communities of Practice&lt;/a&gt;: find where experts in your field hang out, curate ruthlessly, and absorb the vocabulary.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://www.sloww.co/ultralearning-book/&#34;&gt;Project-Based Learning&lt;/a&gt;: solve a real problem &lt;em&gt;you&lt;/em&gt; have, fail + learn + iterate, with other people.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Ancient wisdom (Stoicism, Buddhism, Confucianism, etc.) broadly aligns, but there are a few differences.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;&lt;strong&gt;Slow vs fast&lt;/strong&gt;. Ancient wisdom suggests that judgement &lt;em&gt;must&lt;/em&gt; develop slowly. Science is optimistic about acceleration, e.g. perceptual learning, simulation, etc.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Moralility&lt;/strong&gt;. Ancient wisdom anchors judgement in morality. Science is more agnostic.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;Sometimes &lt;a href=&#34;https://www.cambridge.org/core/journals/bjhs-themes/article/epistemic-demarcations-as-social-erasures-taste-and-the-politics-of-distinction-from-the-revolutions-of-wisdom-to-the-green-revolution/FF24FC14ECA4D30BF4E819D34268FAC8&#34;&gt;&amp;ldquo;taste&amp;rdquo; is just &amp;ldquo;elitism&amp;rdquo;&lt;/a&gt;. That&amp;rsquo;s not going away, and offers another way to develop taste: via &amp;ldquo;club membership&amp;rdquo;.&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;AI, like &lt;a href=&#34;https://link.springer.com/article/10.1007/s00146-025-02422-7&#34;&gt;most automation, erodes skills&lt;/a&gt;. This has happened in the past and we deal with it differently.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Autopilots&lt;/strong&gt; eroded flying skills - which is dangerous. So we &lt;strong&gt;enforce&lt;/strong&gt; flight simulators. Same for surgical knots (robotic surgery), celestial navigation (navy), manual dosing (nurses).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Spreadsheets&lt;/strong&gt; eroded calculation skills. We &lt;strong&gt;leveled-up&lt;/strong&gt; from sums to strategy. Same for CAD, electronic trading, spell-check.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Photography&lt;/strong&gt; eroded painting skills. We &lt;strong&gt;switched&lt;/strong&gt; value to impressionism, cubism, etc. Same for vinyl records, luxury watches, craft coffee.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;GPS&lt;/strong&gt; eroded navigation skills. We &lt;strong&gt;accepted&lt;/strong&gt; this and don&amp;rsquo;t care much. Same for phone numbers, spelling, mental maths.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Think about how the skill we lose will evolve. Then enforce, level-up, switch, or accept accordingly.&lt;/p&gt;
&lt;hr&gt;
&lt;!--
https://claude.ai/chat/859b4fe8-0ad4-460b-87e4-8643a31ea973
https://gemini.google.com/app/6ef791596112da80

---

Comment I wrote at https://www.linkedin.com/feed/update/urn:li:activity:7415616572653744128/?dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287415711348979863552%2Curn%3Ali%3Aactivity%3A7415616572653744128%29

History gives us a few ideas.

Autopilots reduced demand for junior pilots. Regulators now mandate manual flying hours and simulators.

Chess engines made studying pointless since phones crush grandmasters. Instead, engines became training partners and today&#39;s young grandmasters are stronger than ever.

Spreadsheets eliminated the manual ledger work that taught accountants where errors hide. We redefined expertise from exhaustive checking to designing tests and understanding systems.

Surgical robots reduced the hands-on reps that train junior surgeons. Simulation centers and graduated autonomy models now decompose skills into separately trainable components.

No idea which way we will go with software, but I do hope it is the chess route, where junior developers kick the seniors&#39; ...

--&gt;
&lt;p&gt;Source: I used &lt;a href=&#34;https://claude.ai/public/artifacts/1ab0169b-566f-4f2c-b94c-a5fa95838dfc&#34;&gt;Claude Deep Research&lt;/a&gt; and asked &lt;a href=&#34;https://gemini.google.com/share/ad78f0536411&#34;&gt;Gemini to interpret it&lt;/a&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;p&gt;&lt;strong&gt;23 Mar 2026&lt;/strong&gt;. &lt;a href=&#34;https://mtrajan.substack.com/p/leverage-in-the-ai-age-is-not-what&#34;&gt;Taste may not matter as much as I thought&lt;/a&gt;. I see AI learning &amp;amp; acquiring good taste in code (e.g. architecture) and art (e.g. writing, visual design). Accountability may be more important.&lt;/p&gt;
</description>
    </item>
    <item>
      <title>Hot cookies</title>
      <link>https://www.s-anand.net/blog/hot-cookies/</link>
      <pubDate>Sat, 31 Jan 2026 20:36:03 +0800</pubDate>
      <guid>https://www.s-anand.net/blog/hot-cookies/</guid>
      <description>&lt;p&gt;I ordered a &lt;a href=&#34;https://www.cookiemanindia.com/products/caramel-cashew-cookies-soft-chewy&#34;&gt;Caramel Cashew Cookie - Soft &amp;amp; Chewy&lt;/a&gt; at the Chennai airport, an hour before my flight.&lt;/p&gt;
&lt;p&gt;&lt;img loading=&#34;lazy&#34; src=&#34;https://files.s-anand.net/images/2026-01-31-hot-cookies.avif&#34;&gt;&lt;/p&gt;
&lt;p&gt;I&amp;rsquo;ve had cookies before, but not &lt;em&gt;heated&lt;/em&gt;. The person at the counter put it in the microwave for 30 seconds before handing it to me.&lt;/p&gt;
&lt;p&gt;It was the best discovery I made in Jan 2026!&lt;/p&gt;
&lt;p&gt;It is crumbly. It is chewy. It melts. It soaks. It bends when you pick it up. It&amp;rsquo;s warm. It&amp;rsquo;s sweet. It&amp;rsquo;s nutty. It&amp;rsquo;s gooey. Oh, I could go on. It&amp;rsquo;s heavenly.&lt;/p&gt;
&lt;p&gt;Since then, I&amp;rsquo;ve had a few cookies - all heated. They &lt;em&gt;always&lt;/em&gt; taste better hot - especially if they&amp;rsquo;re almost melting. (The taste of the &lt;em&gt;more buttery&lt;/em&gt; ones improves more.)&lt;/p&gt;
&lt;p&gt;There&amp;rsquo;s a reason for it.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;a href=&#34;https://link.springer.com/article/10.1186/s13411-015-0040-2&#34;&gt;~80% of taste is actually smell&lt;/a&gt;. When heated, cookies vaporize carrying the smell of vanilla, caramel, butter, etc from the mouth to the nose.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://www.nature.com/articles/nature04248&#34;&gt;TRPM5, a taste &amp;ldquo;valve&amp;rdquo;, opens when hot&lt;/a&gt;. That&amp;rsquo;s why melted ice cream tastes sweeter. So do hot cookies.&lt;/li&gt;
&lt;li&gt;&lt;a href=&#34;https://pubmed.ncbi.nlm.nih.gov/40625310/&#34;&gt;Heat melts fat, which coats the mouth evenly&lt;/a&gt;. That carries the flavors to every corner of the mouth.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Some other foods that taste better hot are:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Brie or Camembert cheese&lt;/li&gt;
&lt;li&gt;Peaches, plums, and nectarines&lt;/li&gt;
&lt;li&gt;Dark chocolate (like in a sizzling brownie sundae or a choco lava cake)&lt;/li&gt;
&lt;li&gt;Banana bread (or any dense loaf / cake)&lt;/li&gt;
&lt;li&gt;Pecan pie&lt;/li&gt;
&lt;li&gt;Cinnamon rolls&lt;/li&gt;
&lt;li&gt;Leftover pizza&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;I haven&amp;rsquo;t tried them all, but I repeat: hot cookies are my best discovery this month!&lt;/p&gt;
&lt;p&gt;Source: &lt;a href=&#34;https://gemini.google.com/share/7b7ab0698c98&#34;&gt;Gemini&lt;/a&gt;&lt;/p&gt;
</description>
    </item>
    <item>
      <title>Touch people over the Internet</title>
      <link>https://www.s-anand.net/blog/touch-people-over-the-internet/</link>
      <pubDate>Wed, 30 Oct 2002 12:00:00 +0000</pubDate>
      <guid>https://www.s-anand.net/blog/touch-people-over-the-internet/</guid>
      <description>&lt;p&gt;You can hear people over the Internet. You can see people over the Internet. Now &lt;a href=&#34;http://news.zdnet.co.uk/story/0,,t269-s2124692,00.html&#34;&gt;you can touch people over the Internet&lt;/a&gt;. When can we smell and taste them? Or are these senses too mundane to worry about? I, for one, would be very happy to be able to taste recipes online. &lt;a href=&#34;http://www.ravikiran.com/?RequestId=59&#34;&gt;via Ravikiran&lt;/a&gt;&lt;/p&gt;
</description>
    </item>
  </channel>
</rss>
