Welcome to The STEM Sessions Podcast. I am your host, Cody Colborn
In this episode I discuss the increased use of science to get clicks on youtube
- And volley some criticism towards the people doing it – especially those deemed to be science communicators
To be clear, I’m not talking about actual experts, either by education or experience, who share their knowledge
- That’s a new form of teaching which I really enjoy
- Instead, I’m talking about people injecting science into their videos to set themselves apart and boost their perceived expertise on the subject at hand
- This is not a good thing, because they so often apply science and engineering irrelevantly, incorrectly, or unnecessarily, which I feel is a net negative to their viewers
For the record, I don’t consider myself a science communicator, lest you think I’m being a hypocrite
- I’m a mechanical engineer and I know my areas of expertise
- While I like learning outside of my expertise and sharing what I’ve learned, I always make it very clear I’m not an expert and ask you to confirm everything I say
- I even do so at work when I’m writing up a lesson learned that required reading research papers or consulting with actual experts
And a bit of non-podcast news: I’m restarting our in-person meetups, so look for the details on our website or meet-up dot com.
All that said, let’s start off year with a bit of ranting. With apologies to Dave Mustaine of Megadeth…
This is The STEM Sessions Podcast Episode 25 – STEM Sells, and We’re Buying
Injecting science into arguments on the internet has always been a way of making the author sound smarter or carry more authority
- On one extreme, we have a subset of edge-lords I like to call STEM-lords did this frequently to rile people up, especially in arguments of religion and politics
- They know just enough to make themselves sound smart and claim victory by pointing out their target’s logical fallacies
- Their less aggressive counterparts are followers of ideas that are politely categorized as pseudo-science such as flat-earth, free energy, and cryptozoology
- They apply a sort of scientific process to their arguments and observations to give them perceived academic credibility
On the other extreme, and perhaps in reaction to the successful public relation campaigns of those pseudo-science circles, people who actually work in the science fields entered the fray
- An early example is found in 2008 when Dr. Steven Novella and Dr. David Gorski popularized the term science-based medicine
- According to their about section, science based medicine was started because “online information about alternative medicine is overwhelmingly credulous and uncritical” which leads to confusion and uninformed consumer decision-making
More and more academics and professionals have recently joined in, especially since the COVID pandemic has seemingly bolstered the perceived authority behind the term science
- You see this a lot in the fitness realm of YouTube with sports medicine scientists countering the “get ripped in only seven minutes a day” gurus
- You also see this in areas like archeology, linguistics, psychology, with Dr. Jackson Crawford and his videos of indo-european languages and norse mythology being a prime example
Those are true experts in their fields
- And while they are ultimately selling a product such as books or classes or themselves, they’re still experts with professional and academic integrity, so I’m less critical of their teachings
But then we have a category of authors that exist between the two extremes
- They are neither experts in STEM nor conmen
- They’re they layperson who has carved out a niche on YouTube and now inject science to get more audience attention
- They use science as clickbait
- And that’s the category I want to discuss further
Science sells in the YouTube algorithm – at least until it’s inevitably tweaked in the near future – so I don’t blame these authors for trying to standout
- Most of them think they are legitimately using science to add value to their topics
- Issue is the science they use is frequently incomplete or irrelevant or sometimes wrong (again, not from malicious intent)
- It’s especially nails on a chalkboard for me when the person who does this is a science communicator of some renown, because you would think such a person would know better
First example is outdoor gear reviewers
- Hiking, camping, and backpacking are hobbies of mine
- Also like collecting gear much to the dismay of my closets, garage, and bank accounts
- So I watch many gear review channels
- Used to be about how the gear feels, how well it’s made, how durable it is
- Most scientific you got was does the advertised weight accurately reflect the actual weight
Now, several channels are all about scientifically testing new pieces of kit
- How fast can a stove or fuel boil water?
- How long will fire starting material burn?
- How much heat does a candle stove produce?
Possibilities for testing are endless, and non-biased testing should be done more often with results readily available to the consumer
- So I applaud this crowd for trying
- And while I believe they believe they are legitimately trying to add value to the conversation, their methods often are not as scientific as they think
- In a water boiling test, for example, they don’t control against convective heat loss or changes in wind by providing a windscreen around the burner
- In a burn duration test, they may not control for moisture in the surface, which would cause the subsequent materials to burn longer because the first material spent energy drying out the stump being used as the testing surface
And in nearly every case, they don’t repeat tests
- How can results be considered scientific when they don’t test for repeatability?
- Even doing three burn tests and calculating the average is better than calling the matter concluded after one trial
That said, there are a handful of channels that perform controlled and value-added testing
- I’ve seen lab quality testing done on portable solar panels and battery packs and rock climbing gear
- But those authors almost always have a background in STEM, usually engineering, and sometimes that’s even their dayjob
- You can definitely tell the difference
Second example is the guy who collects a lot of data, but doesn’t know how to take full advantage of it
- Disney vlogger who routinely talks about his love for the “science of Disneyland”
- Typically involves observing crowd flow and establishing benchmarks in ride queues – i.e. if the line is backed up to point X, then the wait will be 25 minutes
- All great and helpful observations to note and publish
He also studies wait times by going on a ride a bunch of times, noting the posted wait time at the start and comparing it to the actual wait time
- Again, these observations are great
- It’s helpful to know Big Thunder Mountain wait times are typically over-estimated by 12 minutes
- But an abundance of data does not equate to science if you don’t apply the correct statistical analysis
For example, why did he determine 30 runs was a statistically relevant number?
- Could he have reached the same conclusion using 15 runs?
- Or should he have done 60?
- He never explains his reasoning
Why use the mean of the wait times instead of the median?
- Mean is the average which is found by adding together all of the results and dividing by the number of values in the dataset
- Median is found by putting the values in ascending order and locating the middle value
Both return a value somewhere in the middle of dataset, but they can be different values and are applied in different use cases
- Most references will tell you to use mean when you have a normal distribution (also called Gaussian distribution) of data – think of the shape of a bell curve
- Conversely, median is applicable when the distribution is skewed with more values on one end of the range than the other
- Median is applicable when you have a wide spread of data, with extremes on both ends of the set
- Whereas mean works better when the values are all closer together
Let’s say you have a five value set (to make the math easy) of 5, 7, 7, 18, and 33
- Mean is 14
- Median is 7
- That 33 is an outlier in the data set
- If removed, the mean becomes 9.3 and the median stays at 7
However, you can’t throw out data “just because”
- You need to justify its exclusion
- Was it instrumentation error?
- Was an external force at play which wasn’t present in the other runs?
In the case of the Disney vlogger, his dataset produced a mean of 42 minutes, but the median is 30 minutes – difference of 12 minutes or 30-some%
- Data ranged from 5 to 94 minutes, with 30 entries
- Many factors that impact queue lengths in an amusement park are difficult if not impossible to control for
- So I think the vlogger is correct in not excluding any of the outliers
- However, using the mean is not the optimal way to summarize this dataset
- He should use median, if not more complex analysis
- In grand scheme of things, 12 minutes isn’t a big deal, but if you’re going to advertise “science”, let’s use the appropriate tools
And that brings us to the last example I want to discuss: people who should know better, but still use science for clickbait
- These are the science communicators on YouTube and other platforms
A few months ago, the channel “Adam Savage’s Tested” published a video titled “Why is Apple’s USB-C Cable $130?”
- Thumbnail image is Adam holding up two cables side by side with the text “$13 vs $130”
- As a mechanical engineer, I assumed I was about to watch a side by side of comparison of cables with different price points but still meet the same goals
- Instead, what I watched was a video sponsored by a company called Lumafield showcasing its scanning technology
Scanned the internal constructions of the following cables
- $130 Apple Thunderbolt 4 cable
- $11 Amazon Basics USB-C
- $5.50 generic USB-C
- $3.90 generic USB-C
Comparing USB-C cables to any Thunderbolt cable, let alone Apple’s, in an attempt to justify the latter’s high price tag is disingenuous at best, and total BS propaganda at worst
- Thunderbolt and USB-C cables are not the same product at all
- Wildly different use cases
- Of course, the Thunderbolt cable is going to be better engineered and constructed
- It needs to be to meet a requirement spec well beyond a the requirements of a basic USB-C cable
- More appropriate comparisons would be less expensive Thunderbolt cables to Apple’s, or cheap USB-C to expensive USB-C
Lest you think I’m being needlessly picky, in the video’s introduction, Adam himself says, “we have scans of Apple cables and their imitators”
- That’s simply deceitful – a USB-C cable is not a Thunderbolt imitator
- They’re not even apples to oranges in comparison
Why didn’t they compare the $130 Apple Thunderbolt cable to a $70 Thunderbolt cable from Belkin or the $40 Thunderbolt cable from Anker – I see both on Amazon as I’m writing this
- That would have been an honest comparison
- Instead, we get 22-minutes of three guys ogling the design and solder joints of an Apple thunderbolt cable, and justifying its price tag by showing how junky the cheap USB-C cables in comparison
- Maybe more importantly, it’s 22-minutes of marketing the heck out of the Lumafield scanners
- That’s a misuse of science and engineering, especially coming from a channel proud of its science communication
- Fortunately, many of the video comments saw through the farce
This episode isn’t meant to dunk on these channels – at least not fully
- It’s meant to highlight that using science as a differentiator doesn’t always make what’s being discussed better or carry more weight
- More and more, science and science communicators are being placed on pedestals
- And as a result, we see science successfully used in marketing and propaganda
- Apparently, just like sex, science sells
Thank you for listening to this episode of The STEM Sessions Podcast; researched, written, and produced by Cody Colborn. Shownotes can be found at thestemsessions.com. Feedback and corrections are always welcome.
If you received value from this episode, and wish to give some back, please visit thestemsessions.com/valueforvalue for ways to support the podcast.
Please remember, STEM belongs to everyone. We should not allow it to be siloed or gate-kept by experts, policy makers, or talking heads. Bias is found in every message, so always verify what you read and what you’re told.
Until the next episode, stay curious.
REFERENCES
https://sciencebasedmedicine.org/about-science-based-medicine/
https://www.cuemath.com/data/difference-between-mean-and-median/