diff --git a/instructors/instructor-notes.md b/instructors/instructor-notes.md
index 44f4c61..36226eb 100644
--- a/instructors/instructor-notes.md
+++ b/instructors/instructor-notes.md
@@ -93,12 +93,27 @@ evaluation feature of Snakemake, until we are ready to properly introduce and un
 
 ## Episode 03 - Chaining rules
 
+### Illustrating the wildcard matching process
+
 There is a figure to illustrate the way Snakemake finds rules by wildcard matching and then tracks
 back until it runs out of rule matches and finds a file that it already has. You may find that
 showing an animated version of this is helpful, in which case
 [there are some slides here](
 https://github.com/carpentries-incubator/snakemake-novice-bioinformatics/files/9299078/wildcard_demo.pptx).
 
+### Named inputs versus lists of inputs
+
+In this course, we introduce named inputs and outputs before lists of inputs and outputs. This
+results in shell commands like:
+
+`"kallisto quant -i {input.index} -o kallisto.{wildcards.sample} {input.fq1} {input.fq2}"`
 
+Rather than the less readable version with a simple list of inputs:
 
+`"kallisto quant -i {input[0]} -o kallisto.{wildcards.sample} {input[1]} {input[2]}"`
 
+Later, we introduce lists of inputs in tandem with the `expand()` function. Of course it is
+possible to have a list of outputs, but this is uncommon and not needed to solve any of the
+challenges in this course. In fact, introducing lists of outputs may confuse learners as they
+may think it is possible for a rule to yield a variable number of outputs in the manner of the old
+`dynamic()` behaviour, which is not a thing.
diff --git a/learners/prereqs.md b/learners/prereqs.md
index b3495b6..9f6884f 100644
--- a/learners/prereqs.md
+++ b/learners/prereqs.md
@@ -58,18 +58,18 @@ $ ls -ltrF --directory --human-readable foo_*
 * Know how to look up the meaning of such options in the `man` page
 * See that `foo_*` is a wildcard (aka. glob) pattern that may match multiple file names
 
-And these specific tasks occur within the course:
+And these specific concepts occur within the course:
 
 * Use `cd` to change the current working directory
 * Use `&` to run a command as a background job
-* Use `>` and `<` to redirect output to, and input from files
+* Use `>` and `<` to redirect terminal output to, and input from, files
 * Use `|` to redirect (pipe) output directly between commands
 * Use `wget` to fetch a file from a web URL
 * Create and remove directories with `mkdir` and `rmdir`
 * Use `rm` and `rm -r` to remove files and non-empty directories
 * Use `ln -s` to create a symbolic link
 * Use `cat` and `head` to show the contents of a text file in the terminal
-* Run a Bash script file containing shell commands
+* Run a Bash script file containing shell commands: `bash scriptname.sh`
 
 ## Biology and bioinformatics
 
@@ -87,14 +87,14 @@ overlap, but more typically the fragments loaded in the machine are longer, and
 middle of the fragment are never read.
 
 The data from the machine is saved into a file format named [FASTQ](
-https://en.wikipedia.org/wiki/FASTQ_format) which contains both the sequence (in *ATCG* letters)
+https://en.wikipedia.org/wiki/FASTQ_format) which contains both the sequence (of *ATCG* letters)
 and the per-base quality score, which is an estimate of the error rate recorded by the machine as
-it runs. Average quality drops off as the sequencer reads further into the fragment.  It is
-generally desirable to discard low-quality data, either by trimming off bad bases or discarding
-the whole read.
+it runs. Average quality drops off as the sequencer reads further into the fragment. It is
+generally desirable to discard low-quality data, either by trimming off bad bases from the end or
+discarding the whole read.
 
 In the most common case, the next step in DNA analysis after quality filtering is to map the
-reads on a known reference genome, which is the job of [an aligner](
+reads onto a known reference genome, which is the job of [an aligner](
 https://en.wikipedia.org/wiki/List_of_sequence_alignment_software#Short-read_sequence_alignment).
 This finds the most likely location where the raw read came from, allowing for a certain degree
 of sequence mismatch. The second most common idea is to perform a [de-novo assembly](