m-chrzan.xyz
aboutsummaryrefslogtreecommitdiff
path: root/src/blog
diff options
context:
space:
mode:
authorMarcin Chrzanowski <m@m-chrzan.xyz>2021-03-16 16:33:53 +0100
committerMarcin Chrzanowski <m@m-chrzan.xyz>2021-03-16 16:33:53 +0100
commitaf815c1e87dc0185f9167365b01290050bbc110f (patch)
treeb4604dc276e27d5823f29ace8875f8b2f6b7d105 /src/blog
parent5e45c065b0f0e7cc94a979cfea7677028af801cc (diff)
Publish Nothing is Good for You
Diffstat (limited to 'src/blog')
-rw-r--r--src/blog/nothing-is-good-for-you.html41
1 files changed, 41 insertions, 0 deletions
diff --git a/src/blog/nothing-is-good-for-you.html b/src/blog/nothing-is-good-for-you.html
new file mode 100644
index 0000000..7a6ef18
--- /dev/null
+++ b/src/blog/nothing-is-good-for-you.html
@@ -0,0 +1,41 @@
+title: Nothing is "Good for You"
+date: March 16, 2021 15:06
+---
+<p>
+Popular science websites or lifestyle bloggers often tout headlines along the
+lines of "Is X good for you? Science says YES!" These articles usually cite some
+new study, though without going into the details of it, rarely discussing even
+the surface level of the methodology or significance of results.
+</p>
+
+<p>
+The study measures the impact of X on some output(s) Y, and just based on the
+fact that it concludes with a positive impact on Y (no matter the magnitude),
+the pop science article reports that X is objectively good, and will end by
+recommending that every reader should now do X, incorporate X into your daily
+routine, need never feel guilty again doing X (since often X is something
+associated with bad habits, like drinking a particular alcohol, playing video
+games, etc).
+</p>
+
+<p>
+Clickbaity headlines like that will easily imprint a positive association with X
+in most readers' minds. But just because a study found a positive impact of X on
+Y doesn't necessarily warrant saying that it is objectively "good for you".
+</p>
+
+<p>
+Even barring general problems with many studies (was it performed on a large
+enough population? was the methodology correct? were there any biases
+introduced? was the population representative of the reader?), any study will be
+able to test only a limited number of outputs Y. So even if the study was
+performed perfectly, and you agree with the authors on what change in Y is
+positive or not, the study can make no claim on all other aspects that could be
+potentially impacted by X, including ones that we wouldn't even expect to be
+affected.
+</p>
+
+<p>
+You will most likely benefit more from using "Lindy" things (per Nassim Taleb's
+terminology) than jumping on any particular pop science trend.
+</p>