Here's something that happens to me all the time: a student can solve a problem one day. The next day, or a few days later, I give them a different problem on the same topic and they don't know what to do. What happened? Why didn't the learning stick?

Forgetting is maybe the most ubiquitous experience in education. Teachers teach, sometimes students learn and sometimes they don’t. Learning is a bit mysterious. There's no magic formula to make sure students remember everything they learn. Still, I think it's helpful to lay out a few things that improve long-term retention.

One common reason why learning doesn’t stick is that students don’t practice enough. That’s certainly often the case. Practice will be one idea I talk about below. But I worry that “students need more practice to remember things” becomes a narrow and limiting mental model. It’s easy to end up in a trap where everything is about practice, and the only response to students struggling is asking them to practice more. It’s also easy to give students repetitive, poorly designed practice that doesn’t help much. That’s the brute force approach to memory, and while practice is important it needs to be paired with a few other elements so that learning lasts.

Thinking

I realized recently that I didn't know how many steps there are going up to my front door. It could be 2, or 3, or 4, or 5. I can picture the front of my house, I can imagine myself walking up those steps, but I couldn't tell you how many there are. I've walked into my house thousands of times. But I don't typically think about the number of steps, so I don't remember it.

There's a difference between doing something and learning from it. The key question to ask in school is, "what are students thinking about?" Students could solve a hundred problems about finding circumference but learn nothing. If the whole time they're just thinking "I answer these problems by multiplying the number in the problem by 3.14," that practice won't do any good because they aren’t thinking about circumference. Just as I walk into my house every day without thinking about how many steps there are, many students go through the motions without thinking about the mathematical ideas we want them to learn.

Here are a few things to consider about thinking:

• Students learn what they think about. As often as possible I want students thinking "circumference is equal to pi times diameter" and not just thinking “multiply by 3.14.”

• Students should be thinking about the deep structure of the concept as often as possible. Circumference isn't just pi times diameter, it's also 2 times the radius times pi, and circumference divided by pi is diameter, and circumference is proportional to diameter. All of those ideas form one connected network rather than a single procedure. I want students thinking about the connections in that network.

• Don’t overload students. Our working memories are limited. If we ask students to think about too many different things at once, their working memories become overwhelmed and they won’t learn much at all. Focus on one thing at a time. This also means students need fluency with all the little pieces of a bigger problem, so those little pieces don’t take up space in working memory that should be dedicated to learning new ideas.

• All the above assumes students have the bandwidth to be thinking about math. If they're feeling anxious or distracted by a Snapchat they just got or preoccupied by something happening outside school or the room is chaotic, they are less likely to be thinking about math. We should try to eliminate those distractions as often as we can.

Connect to What Students Already Know

Teaching students to combine like terms can be rough. Some students seem to get it right away, and others are perpetually confused. They’ll tell me 2x + 3x = 5, and just as they figure out that it’s 5x I ask them 8x - 4x and they tell me 4 — they’re supposed to subtract the x’s, right? And then we get to 8x - x, and that one is definitely 8, right? And 2x + 5 + 4x is a complete disaster. I eventually stumbled across a really nice metaphor. We start with m’s. M stands for a million. What’s 2m + 3m? Well 2 million + 3 million is 5 million. 8m - m is 7m, because 8 million - 1 million is 7 million. Then we expand out to billions, thousands, and ones. The key idea here is to take what we’re learning, and connect it to something students already understand. Learning sticks much more easily when students build connections between ideas and link new learning to prior knowledge.

This applies to all sorts of topics. We learn new things in relation to what we already know. Division is easier to learn if you have a solid grasp on multiplication. Proportions are easier to learn if you build on intuition for everyday proportional relationships like cost and speed. Operations with negatives build on familiar rules with positive numbers.

Here are a few specific ways to do this:

• Make sure students have the prior knowledge to build on. Check to make sure students have the prerequisites they need, and reteach if they don’t.

• Before introducing a new idea, ask a bunch of review questions on related ideas to get that prior knowledge fresh in students’ minds. This isn’t random review, and it isn’t necessarily what the class learned yesterday — the goal is to tee up the connections we want students to make.

• Make the connections between new ideas and prior knowledge clear. Get students thinking about the ways that new learning builds on what they already know. Ask students how ideas are connected. Don’t take those connections for granted.

• Don’t just make the connection between new learning and prior knowledge once. Reiterate those connections multiple times, over multiple days.

Retrieval Practice

For a long time I couldn't remember the difference between affect and effect. Every so often I would need to use one of the words. I would look it up, use the right word, and move on with my life. A few weeks or months later I would need to use one of the words again, realize I'd forgotten, and look it up again. One reason the difference didn't stick is I was never retrieving what I knew from memory. Each time I was looking up the difference, rather than pulling it out of my memory. Retrieval is better for retention than rereading something or trying to learn it through pure repetition. I was looking up the difference between affect and effect, but never retrieving it. Eventually I got tired of forgetting and set some reminders to quiz myself about affect/effect every day or two. That retrieval helped, and now I don't need to look up the difference anymore!

I might have a really clever way to introduce solving equations, connecting it to what students already know and getting students thinking about the structure of equation-solving. All of that focuses on how the information enters the brain. If students never pull that information out, the learning won’t stick. That’s what retrieval practice is all about.

Here are a few things to consider when structuring retrieval practice:

• Do it! For a long time I just didn’t make retrieval practice a priority. It doesn’t need to be a ton of practice every day, short chunks of regular retrieval can be really helpful.

• Space it out. Retrieval practice should happen the day students learn something new, the next day, the next week, and the next month. If students forget, reduce the interval and start again.

• Interleave. When different topics are interleaved together students need to figure out which ideas to retrieve, preventing them from going on autopilot.

• Make sure students are actually retrieving. If you ask a question and students don’t remember it, that means they could use a quick reteach and then another round of retrieval soon.

• Retrieve in a variety of ways. Ask different types of questions, and use different contexts.

• Avoid preempting retrieval. If I want students to remember something they need to remember it themselves. Giving students too many hints to start practice or putting the steps on the board can short-circuit retrieval practice and hurt retention.

Transfer

Here’s a problem: a bat and a ball cost $1.10 together. The bat costs $1.00 more than the ball. How much does the ball cost? (Answer in footnote.)1 I have solved lots of math problems like this one, but I got this one wrong when I first saw it.

The problem above is designed to be tricky. I don’t feel too bad about getting it wrong. But that’s how students feel a lot of the time: they learn one thing, and as soon as they feel comfortable with it we start asking them in lots of different ways that seem designed to be confusing. A student can solve a Pythagorean Theorem problem when there’s just one triangle, but once we make the diagram more complex they get stuck. We don’t want students to only be able to solve problems when we ask them in predictable ways, we want them to be able to apply that knowledge in lots and lots of different contexts. This is the toughest part of learning. There are no shortcuts, there’s no magic sauce that will help me see forks in chess games or help students realize that a word problem can be solved with a system of equations. It’s a slow, gradual process of practicing skills in new contexts and thinking about the deep structure of a mathematical concept.

Here are a few ideas:

• Use lots of examples and non-examples. Explanations are overrated. Students learn best when they see a lot of examples of how an idea applies and think about what those examples have in common. Students also need non-examples to understand where an idea doesn’t apply.

• Variety is key. Start with more predictable question types to help students gain confidence, but as soon as they’re ready for it ask questions in different ways and in different contexts.

• Focus on the deep structure. If you want students to apply systems of equations to lots of different situations, get students thinking about the deep structure of a system of equations: two or more different constraints that can be represented as equations. Focusing on that deep structure makes it more likely students can apply their learning in the future.

• Avoid too much repetitive practice. Repetition is helpful early on in the learning process, but as soon as students are ready for it practice should be varied and interleave different topics.

• Be patient. Transfer is hard. Don’t give up, keep trying.

Final Thought

There’s no flowchart to figure out which of these four ideas is the cause of a particular student forgetting. And I have a class full of students who are probably forgetting for different reasons. Still, with every topic I teach, I can point to places where I can improve. I’m not a perfect teacher. When students aren’t remembering something I want them to remember, I use these four ideas as a place to start. I can always find something to tweak that will help students remember.

That’s the heart of this post. Forgetting and remembering aren’t random. They’re complex, and it’s never a sure thing. But there’s always something I can improve. If students aren’t remembering, it’s my job as the teacher to make a change. Get students thinking about meaning. Prompt connections to prior knowledge. Design effective retrieval practice. Ask students to apply what they know many different ways. Teaching doesn’t do much good if students don’t remember what they learn.

Sometimes I find an old game and it makes me notice or wonder something mathematical. For example, I found Rubik’s Tangle 2 in a charity shop.

In his memoir, Rubik describes a period after the initial cube craze died down where he worked on other puzzles, including this one. My copy of Tangle 2 is dated 1990, placing it within the second wave of the Rubik’s cube. On the box he explains that he hopes to “create things that challenge the mind—which everyone can play and enjoy”.

Tangle is a tile-based puzzle where the aim is to arrange the tiles into a square so that the ropes form continuous lines. You can see some example tiles in the header image.

At first I thought it would be quite complicated to work out how many tiles you could make like this. There are eight points around the edge of the square tile, two per edge, where four-coloured ropes enter and leave the tiles.

After a little examination, though, I realised the same asymmetric design is on all cards, and every card uses each colour once. This is a much simpler counting problem! There are four options for the colour of the first rope, then three for the second, two for the third, and the remaining colour goes to the fourth. There are $4!=4 \times 3 \times 2 \times 1 = 24$ possible tiles. The game comes with 25 tiles so you can make a square. Since we have 24 different patterns and 25 tiles, it follows that there must be more than one tile for at least one of the patterns (a concept called the pigeonhole principle).

In fact, all of the 24 different patterns are included and exactly one is duplicated; the duplicate tiles are these:

Rubik explains that they made four variants of the Tangle puzzle. Each had the same basic 24 tiles with a different one duplicated. As an added bonus, if you combine all four sets, you can find a solution for them all together which fills a $10 \times 10$ square. Also, he says with the basic set you can find one solution to cover the surface of a $2 \times 2 \times 2$ cube so that there are four endless loops in the four colours. There are games with tiles that are harder to count. My next charity shop find is Squominos, a tie-in game for the 2005 Charlie and the Chocolate Factory film. Apart from three Wonka wild cards, each tile features four Oompa-Loompas in different colours and the game sees players placing tiles so that adjacent Oompa-Loompas match in colour:

Despite the colour choice similarity, these tiles are different from those in Rubik’s Tangle in two important ways: the design is symmetrical, and each colour may be used multiple times on a tile.

First let’s deal with the fact we can repeat colours. We now have many more than $4!$ ways to colour a tile. We can choose from four colours for the first Oompa-Loompa, then we have a choice of four for the second Oompa-Loompa too, and four colours for the third and fourth also. There are a total of $4^4=256$ colourings.

The symmetric design introduces a wrinkle at this point. One of our 256 options is when we choose yellow for the first colour, and blue for the other three. Another is when we choose yellow for the second colour, and blue for the other three. And there are two more selections with one yellow and three blue Oompa-Loompas. Using a simplified tile design with each Oompa-Loompa represented by a quadrant of the square, these four tiles look like this:

The thing is, if you hold a physical tile and turn it 90°, you aren’t suddenly holding a different tile. We have counted this design four times when in fact it only represents one distinct tile.

Since there are four tiles that are all the same (the original and one for each of three rotations), we might simply divide $256/4 = 64$ to correct for this duplication. We now have reasonable upper and lower bounds, but we do not have the answer. This is because not all tiles were counted four times in our 256. For example, one of the 256 is the selection where we choose blue for the first Oompa-Loompa, and blue for the others. This is only counted once among our 256:

 

 

Meanwhile, the tile where blue is chosen for the first and third and yellow for the second and fourth is duplicated only once in our count, when yellow is chosen for the first and third and blue for the second and fourth:

What we want to do is divide our count by four only for those tiles where there are more than one in the count that aren’t in fact distinct tiles. One way to do this is to divide 256 by four and then add back in the tiles we have accidentally removed from the count. Another way is to first increase 256 to account for the tiles that weren’t counted four times, then when we divide by four we will obtain the correct number of distinct tiles.

We can do this by considering when applying a symmetry to a tile leaves it looking identical. If this happens, we have a tile that hasn’t been counted four times, so we add one to our total before we divide. There is a clever way to do this using permutations. We can label the tile and consider permutations of the labels:

A permutation is a function which rearranges (or relabels) a set of objects. For example, a rotation of 90° clockwise ($\rho$) changes the tile above to this:

 

 

 

Here the region labelled 1 has been moved by the rotation to where we previously had 2, meanwhile 2 has moved to 3, 3 to 4, and 4 to 1. We can write a permutation as a product of disjoint cycles. In general, $(\dots xy \dots)$ means $x$ shifts to $y$ and $(x \dots y)$ means $y$ shifts to $x$. Our 90° rotation, therefore, is a permutation $(1234)$.

We will use permutations to count the cycles of quadrant labels that must be coloured the same for a tile to be unaffected by this permutation.

The permutation $(1234)$ is a cycle of length 4 because the four elements shift between themselves and if we do it four times we get back to where we started. We’ll label a cycle of length $i$ with $x_i$, so this is an $x_4$ cycle. Let’s collect the information we know about $\rho$:

We can do the same for a double rotation ($\rho^2$). This time, 1 has changed to 3 and 3 has changed to 1, a cycle of length 2. Meanwhile, 2 has become 4, and 4 has become 2, another $x_2$:

A triple rotation ($\rho^3$) is similar to a single rotation, just in the other direction:

The other symmetry we should consider might not seem like it’s worth considering at all: doing nothing. Or, equivalently, rotating four quarter turns. Either way, every element stays where it started. This is the identity ($e$) and is four one-cycles:

Summing our cycle representations, we get $x_1^4 + x_2^2 + 2x_4$. This expression represents the colourings that are unchanged by each symmetry. We can use this to obtain a big number, which if we divide by the number of symmetries will give us the distinct number of tiles. This is called the cycle index and is written

\[ P(x_1,x_2,x_3,x_4) = \frac{1}{4} \left( x_1^4 + x_2^2 + 2x_4 \right)\text{.} \]

Each cycle represents a set of quadrants that must be coloured the same if the tile is to be unchanged by this symmetry. Therefore we can colour each cycle four ways, and the total number of ways to colour the tiles is

\[P(4,4,4,4) = \frac{1}{4} \left( 4^4 + 4^2 + 2 \times 4 \right) = 70\text{.} \]

In this function, each colouring of our tile is included once for each symmetry that leaves it unaffected. Our all-blue tile is included by $e$ and by $\rho$, $\rho^2$, and $\rho^3$, ensuring it is overcounted sufficiently before we divide by four. Our blue, yellow, blue, yellow tile is included once by $e$ and once by $\rho^2$, as is its partner the yellow, blue, yellow, blue tile. Meanwhile, each of the four with one yellow and three blue quadrants is counted by $e$ and not by the other symmetries. In each case, we make sure we are getting exactly four of each distinct tile before dividing by four to get the true number of different tiles.

So, are there 70 tiles are in the Squominos box? Ah, no, there are 55.

What has happened? To find out, really we need a list of the 70 colourings to compare with the 55 tiles.

Actually, I’ve undersold our function $P(x_1,x_2,x_3,x_4)$. It doesn’t only count the number of tiles, it can also tell us how those tiles are coloured.

We can count things by expanding algebraic expressions. For example, if I toss a coin it might land heads ($H$) or tails ($T$). If I represent these choices with an expression$H+T$, I can find the possible outcomes from $n$ coin tosses by expanding $(H+T)^n$. For example, the outcomes from tossing three coins are given by

\[ (H+T)^3 = H^3 + 3H^2T + 3HT^2 + T^3 \text{.}\]

The $H^2$ and $H^3$ represent throwing double and triple heads, respectively. Terms with a coefficient other than one are telling us there are multiple ways to arrange those outcomes, for example $3H^2T$ tells us there are three ways to arrange two heads and one tail: $HHT$, $HTH$, and $THH$.

Here, we can colour a cycle of length $i$, $x_i$, with either $i$ red Oompa-Loompas, or $i$ green Oompa-Loompas, or $i$ blue or $i$ yellow. This means the options for each $x_i$ can be expressed as $r^i+g^i+b^i+y^i$.

We can therefore find all the different colourings using

Each term here represents a way to colour the tiles. Those with coefficients other than one reveal that there are multiple ways to colour tiles with that selection of colours. For example, the coefficient $2$ in the term $2b^2g^2$ tells us there are two fundamentally different ways to colour with two blue and two green Oompa-Loompas. All others are rotations of these:

So what is in the box and why are we missing 15 tiles? Examining the tiles I noticed a couple of omissions. First, consider tiles with two quadrants of one colour and the other two colours different, like $3gb^2y$ which can be coloured three ways:

Only the first two of these are in the box, the third is absent. There are 12 such terms in our expansion, so that accounts for 12 of the missing tiles.

In addition, there are 6 fundamentally different ways to colour the tiles using each colour once ($6bgry$), and only the first three of these are in the box:

At first I noticed that the missing $gb^2y$ is a reflection of one of those included. We didn’t include reflections in our calculations because the tiles are opaque—if you flip a tile over, you are not looking at a reflection of the Oompa-Loompas but the back of the tile. But in situations where the tiles are transparent, you could do the same analysis taking account of the four reflections as well as the rotations. If you do this, you will get this alternative cycle index,

\[ P_r(x_1,x_2,x_3,x_4) = \frac{1}{8} \left( x_1^4 + 3x_2^2 + 2x_4 + 2x_1^2x_2 \right)\text{,} \]

and it happens that $P_r(4,4,4,4) = 55$.

So did the folks behind Squominos do the maths taking account of reflections? I think a couple of bits of evidence point against this.

First, while the missing tiles in the terms like $3gb^2y$ are all mirrors of a tile that is included, we can’t say the same for the three from $6bgry$—two of the three in the box are reflections of each other.

Another strike against clever maths counting is that there is only one tile like $g^2ry$, and one like $bg^2r$ appears twice in the set:

There may be deep gameplay reasons why the manufacturers chose a particular selection of 54 tiles from the 70 possible tiles and then duplicated one. It’s also possible the process of designing the tiles was more heuristic.

One puzzle to leave you to ponder arises from three-colouring our tiles with $P(3,3,3,3)=24$ tiles. You can make yourself a set of these tiles by counting each colouring of $x_i$ using $g^i+b^i+y^i$ and expanding
$P(g+b+y,g^2+b^2+y^2,g^3+b^3+y^3,g^4+b^4+y^4)$. These are a generalisation of dominoes called the MacMahon squares.

Martin Gardner describes what he calls a “first rate puzzle” with the MacMahon squares: fit together all 24 squares in a four-by-six rectangle meeting two conditions: each pair of touching edges must be the same colour, and the outer border of the rectangle must be all one colour. He claimed there was only one solution when he first published the puzzle in his long-running Mathematical Games column in \textit{Scientific American} in March 1961, and later said this “proved to be the greatest understatement ever made in the column”.

Because of the missing tiles, I can’t solve this puzzle using my Squominos tiles, but why not try to make yourself a set of MacMahon squares? So you can check your answer, I’ve printed a set below. But don’t get any ideas about cutting up your copy of Chalkdust—the editors might ban me from ever writing again.

The post How many tiles does this game have? appeared first on Chalkdust.

Our cursed age of self-monitoring and optimisation didn’t start with big tech: as so often, the Victorians are to blame

- by Elena Mary

Read on Aeon

This week is a grab-bag of thoughts on the definition of “conceptual understanding,” recent data on AI and tech usage in schools, and reviews of the recent Frankenstein movie and the not-so-recent Gattaca. So feel free to—cue House of Pain—jump around!1

Teaching for conceptual understanding

The education space is riddled with something I’ve long called “head nodder” phrases, pithy little statements that everyone agrees we should be pursuing with students, yet often lack an agreed upon definition, much less a method of ensuring they are properly learned. Critical thinking! [Heads nod] Growth mindsets! [Heads nod] Creative collaboration! [Heads nod] It all sounds good until you dive into the details and find that people lack a shared mental model of what these phrases really mean.

“Conceptual understanding,” that’s another potential head nodder, but last week my friend and teacher extraordinaire Michael Pershan wrote a great essay on making this idea more concrete—check it out here. His central proposition is that what we really mean by “conceptual understanding” is that students know “true and useful generalizations.” (He adds “about mathematics” because he’s a math teacher, but it applies to other subjects as well I say.) And what I so often love about Pershan’s essays is how he uses real-world examples from his teaching experience to explain what he’s getting after. Consider the following task:

This is a neat little geometry problem, and here’s how it played out one day in Mr. Pershan’s classroom:

When I asked my 7th Graders to answer this, one student pointed out 100% correctly that since it’s four units up from E to D, the diagonal has to be longer than that.

I was about to move on, when I caught myself. I turned back to the kid. “So is the diagonal always going to be more than the vertical distance? Why would that be?”

I’m glad I pushed, because my students responded with two smart generally true answers:

• Yes, the diagonal is always longer than the horizontal or vertical distance, because when going diagonally you’re going in two directions, not just one.

• The Pythagorean Theorem says where the diagonal length is found by adding the squares of the horizontal and vertical distances, and that’s always going to be longer than just the vertical distance squared.

The point of an explanation isn’t just to eliminate doubt—it’s about connecting this particular situation to some generally true fact about the mathematical world.

What I hope you take notice of here is the pedagogical action that Mr. Pershan used here to build conceptual understanding with his students. The probative question he asked was vital to learning something general that gets beyond just getting the right answer. Curriculum cannot teach itself, and the best teaching involves pushing toward this general understanding of a broader principle. (If you think I’m implicitly throwing shade at AI “tutors” here, you’d think right.)

Over in cognitive-science world, the word used to describe the application of general knowledge to solve a novel problems is called transfer. Grant Wiggins, the co-creater of Universal Design by Learning (and who passed away several years ago), described transfer as the central goal of education, the capacity to apply knowledge in new contexts. And while I yield to no one in my advocacy for cognitive science in education, I think my fellow cog-sci enthusiasts sometimes fail to appreciate the vital role that transfer plays in our cognition. As Pershan concludes:

We all want generalization. We want to see how it all fits together. We want to know how a particular idea flows from something bigger. We want to know how to do things—we also want to know things. We want the whole picture. And when we get it—whether we land on it on our own or someone makes us think of it with words or a picture or anything else—the feeling is terrific, like it all finally makes sense.

Yep.

Data on digital device and AI usage in US schools

I miss Dan Meyer. Don’t worry, he’s still around, but he paused his newsletter back in April. That’s been a loss on a number of fronts, but I counted on Dan to keep me informed of the latest data on actual AI usage in schools. Come back to us, Dan!

I’m not going fill the Meyerian void on the regular, but a recent New York Times story about how technology is being used in US schools caught my eye. The main reason being, the reporters actually surveyed teachers—350 of ‘em—rather than relying on one or two anecdotal quotes. To be sure, this is far from a scientific approach to polling, but we try to figure out what’s happening in our education system with the data we have, not necessarily the data we want.2

Here’s what they found:

• 99% of teachers report that their students are provided digital devices in school

• 81% of elementary school teachers said this was true even in kindergarten

• 70% of the teachers who use digital devices in their classroom said they distract from schoolwork, with 36% saying “a lot” and 34% saying “a little”

• 70% of teachers nonetheless said they would continue to use digital devices at least some of the time

• ~66% said student work time on devices had increased since the pandemic

• 64% said digital devices are used in class for standardized testing

• 40% of middle and high school teachers said students spend three or more hours per day on digital devices (!!!)

• And a whopping…6% of teachers…said students use their digital devices for school-approved use of AI (whereas 29% said students are using AI for “non-school activities”)

There’s obviously much we don’t know here, but it’s nonetheless very interesting to me that despite the saturation of schooling with digital devices everywhere, students don’t seem to be using AI all that frequently. Is that because schools are prohibiting them from doing so? Maybe, we don’t know. Is it perhaps because kids don’t like learning by typing things into a box and having a non-human thing generate the statistically most probable text back at them? Maybe, but again we don’t know. What I think we might very tentatively conclude from this data, however, is that there’s a substantial gap between the amount of enthusiasm “in the discourse” around AI’s potential to transform the education system versus the reality on the ground. Warrants mentioning.

The article is also replete with juicy quotes from teachers that I ruefully enjoyed reading—”Kids just want to use A.I. for everything. SO MUCH CHEATING!,” says one high school teacher in Texas—but then there’s this glaring bit of stupidity:

Andy Russell, a product manager at Google who oversees Chromebooks in schools, envisions them ushering in a new era of education—one in which teachers aren’t lecturing in front of the room, but rather acting as facilitators to students learning on computers and using online tools to creatively show what they’ve learned.

“So much of what teaching is today is not the when and the what, like I learned; it’s a lot more of the how—this is how we make a video, this is how we build infographics,” he said.

How many times must a man cite cognitive science, before they will call him a man? How many times must we hear this tired cliché rolled out in support of education technology, a cliché that contradicts the very basic and frankly scientifically indisputable proposition that our ability to understand new ideas depends on the knowledge we have in our heads, what we might call “the when and the what” of basic facts? And while I’m ranting, just how blinkered is Google’s vision of education, or at least Andy Russell’s vision of education, if it’s about making videos and building f’ing infographics? Schools do not exist to create content for YouTube, my dude.

I really think a backlash is growing to all this. I think there are many parents, myself included, who do not want tech-saturated schooling for their kids. There certainly are some in Malden, Massachussettes—check out their letter demanding their local district ban AI. And I’ve started to fantasize about what the alternative might look like. That’s a teaser for a future essay. Maybe.

Frankenstein and Gattaca

Ok, time for two movie mini-reviews. Let’s get weird.

I confess I’ve never read Mary Shelley’s Frankenstein. What’s more, until recently I’d never even watched a movie rendition either. I am sentient human, so of course I’ve been dimly familiar with the basic contours of the tale—crazy scientist creates artificial life by lighting up a blockheaded-looking creature, it gets angry and marches around with its arms out—but this has been a gap in my education.

Into that breech stepped Guillermo del Toro last week with his new version of Frankenstein, now streaming on Netflix. I can’t say I’m a devoted fan of del Toro’s films, I’m not really a goth or horror guy. I did see Pan’s Labyrinth in the theaters decades ago, but all I remember is that weird creature with its eyes in its hands (see above). That said, more recently I’ve appreciated del Toro becoming a very vocal anti-AI advocate who’s been leading film festival crowds in chants of “FUCK AI.” Right on, brother.

But how’s the film? Well…it’s visually stunning, for sure, but I found the first hour pretty ponderous. Victor Frankenstein, the mad scientist, is played by Oscar Isaac—the guy who played Poe Whatever in the newer Star Wars films—and he plays him pretty f’ing mad. Too mad, in my view, it’s a bit over the top. There is, however, one brilliant scene in the first act, where we see Frankenstein prototyping the creature he plans to create, that del Toro breaks down into detail here. I think it’s telling that he eschewed the use of CGI in favor of using actual puppets, actual physical things.

It’s in the second act, when The Creature becomes animated (alive?), that the movie takes off. Jacob Elordi portrays The Creature with such beautiful (and human?) sensitivity, it was impossible to watch without thinking about The Big Important Questions about what it would mean to be something brought into the world with no history, with no like companions, a being of almost unimaginable solitude. Some subsequent investigation reveals that in the novel, The Creature demands that Frankenstein make him a companion, but sadly del Toro chose not to emphasize in his telling of the story; instead, a relationship of sorts develops with a human woman, Elizabeth, that feels a bit forced. Still, the core exploration of what makes us human, what animates our wants and desires, what leads to love, and what happens when this is denied—these feel like the right questions to be asking right now, do they not?

What makes us human sits the center of another great movie I re-watched recently, Gattaca, albeit with a very different lens. I remembered loving this film when it was released in 1997, and it holds up. And makes for a worthy companion watch to Frankenstein, for reasons I’ll try to explain.

You probably know this: Gattaca, directed by Andrew Niccol, is set in a dystopian eugenicist future where one’s life prospects are entirely dictated by ones genes. The genetically “superior,” who have been bred to be that way, live a life of comfort and purpose. The genetically inferior get to clean up after them, literally. The story centers around Vincent Freeman, played by Ethan Hawke, classified as an “in-valid” (get it?) at birth due to his genetic defects, and Jerome Morrow, played by Jude Law, who in contrast is genetically near-perfect—but has been paralyzed by an accident. In order to get into the elite space program, Hawke must pretend to be Law, and go through a series of contorted activities to borrow Law’s blood, his hair, etc, in order to pass as someone who’s genetically fit. In other words, he must borrow another human body. The scenes between Hawke and Law are brilliant, it’s really a love story in some ways (if only they’d gone that far!), and the retro-futuristic film noir cinematography endures beautifully.

Gattaca feels like a time capsule we need to reopen. When the movie was released, The Human Genome Project was in full swing, cloning featured heavily in our monocultural discourse (baa, says Dolly), and it seemed we were on the verge of being able to identify the influence of genes across all our behaviorial traits. Our capacity to become ourselves would be mapped from the beginning. Gattaca of course is a warning against that future, a testament to the horrific dangers of denying our essential humanity through deterministic science.

The future the film imagines future did not come to pass in the short term, but nowadays? “Corporate eugenic companies,” in the words of Eric Turkheimer, are all the rage in Silicon Valley, funded by the same tech billionaires pursuing AI at all costs. They are pursuing the reduction of our human bodies to genetic programming just as rapidly as they are pursuing the reduction of our human cognition to statistical autoregression.

You can connect these dots, I hope, and we’ve covered this ground before. So maybe there is a throughline to this week’s essay. Maybe we need to build greater conceptual understanding of why our essential humanity is worth preserving so that we are better equipped to fight the mounting efforts to degrade it.

The artists are warning us!

Subscribe now

In this article, we will try to understand why OOP is the worst thing that happened to programming, how it became so popular, why experienced Java (C#, C++, etc.) programmers can’t really be considered great engineers, and why code in Java cannot be considered good.

In response to youth suicides, teachers show students the power of headbanging at Fire in the Mountains festival.

The post Heavy metal is healing teens on the Blackfeet Nation appeared first on High Country News.