just drafts

Machine learning in general and large language models in particular are terrific in their potential to solve specific problems, but terrible in how they’re actually put to use. On one end, you have billionaire CEOs from the wealth extractor class who throw insane amounts of money at this technology on the prospect of replacing every worker and employee with machines and make even larger shitloads of money than they already do. On the other end, there are all these people who are nudged into using these models as a new and shiny toy in ways that make them even more distracted from what’s actually going on than they already are. And one of the most popular rationalizations for chatting with an LLM instead of reading a book or doing actual research is the nearly ubiquitous argument of using ChatGPT “only for ideas and inspiration,” not for the actual work.

Which is screaming baloney.

Developing creative ideas—for writing, art, academia, what have you—is already actual work that demands a lot more than feeding prompts to chatbots. That hard work also includes the “structure” of, e.g., an essay, a course, a lecture, a textbook, or a paper. Sure, one can lift such a structure from essays or lectures or textbooks that already exist, which is sort of legit, and that’s literally exactly what you’re doing when you scaffold your work with output from ChatGPT. The other option is to build the structure of an essay or lecture or textbook around an original educational or didactic concept aka a creative idea, which is a lot more work.

On aspects specific to writing, you can read more (with somewhat less snark) over at my primary blog between drafts.

Ed Zitron pulling no punches:

It’s been just over two years and two months since ChatGPT launched, and in that time we’ve seen Large Language Models (LLMs) blossom from a novel concept into one of the most craven cons of the 21st century—a cynical bubble inflated by OpenAI CEO Sam Altman built to sell into an economy run by people that have no concept of labor other than their desperation to exploit or replace it.

Zitron’s been writing about this con game for a long time, particularly its elusive business models and its numbers that are not adding up. The crucial point is that no matter how much generative AI/LLM might be able to change our lives in fundamental ways, there is no product, not even on the horizon. All we get from the major players, while they happily burn through hundreds of billions of dollars, are future promises as absurd as intelligence-insulting on the one hand, and one-day wonders as over-hyped as counterproductive on the other:

Sam Altman believes you are stupid. He believes you are a moron that will slurp up whatever slop he gives you. Deep Research and Operator are both half-products that barely brush against the fabric of their intended purposes, and yet the media screams and applauds him like he’s a gifted child that just successfully tied his shoes.

Among the real major problems of so-called AI is that it will tank the economy with a probability and severity that both grow with every humongous stack of cash set on fire:

Generative AI is a financial, ecological and social time bomb, and I believe that it’s fundamentally damaging the relationship between the tech industry and society, while also shining a glaring, blinding light on the disconnection between the powerful and regular people. The fact that Sam Altman can ship such mediocre software and get more coverage and attention than every meaningful scientific breakthrough of the last five years combined is a sign that our society is sick, our media is broken, and that the tech industry thinks we’re all fucking morons.

This entire bubble has been inflated by hype, and by outright lies by people like Sam Altman and Dario Amodei, their lies perpetuated by a tech media that’s incapable of writing down what’s happening in front of their faces. Altman and Amodei are raising billions and burning our planet based on the idea that their mediocre cloud software products will somehow wake up and automate our entire lives.

The truth is that generative AI is as mediocre as it is destructive, and those pushing it as “the future” that “will change everything” are showing how much contempt they have for the average person.

Thus, welcome to the ultimate hat trick of the twenty-first century: an accelerating climate catastrophe, Western democracies sliding into the hands of Nazi parties and brazen crime lords, and an economic crash that will hit us like a brick the size of the moon.

For the longest time, there were urgent warnings about the kind of Earth we don’t want to leave for our children and grandchildren. These warnings are a thing of the past now, an area of interest for the historian. Today, even if you happen to be an octogenarian, this triple whammy will very likely catch up with you. And for all those of us who are younger than that, in our forties or fifties or sixties even, who’ve felt more or less safe until now from this horrible future we keep creating, sufficiently insulated by life expectancies and our own private modicums of prosperity, it has slowly come closer and closer to become our reality instead, and we’ll soon be able to take a picture.

Addendum
Here’s a follow-up post with a lot of numbers, none of which look good by any stretch.

Among the things that bother me about generative AI is how humans will treat it. But not because generative AI were “conscious” or “alive” in any conceivable way.

Before I make my argument, let me back up a little.

As for not being “conscious” and “alive,” it could be argued that a similar thing once was believed, at the time around Descartes and again in the nineteenth century, about animals—that they were mere automata, and pain for them just a bodily sensation, not something they could actually feel. (What Descartes himself exactly believed is a bit murky when you dig into his writings, but, yeah, that’s probably how he saw it.) This, of course, opened the door for all kinds of animal cruelty that we today as an even more enlightened society have left behi—oh wait. Livestock farming, factory farms, battery cages, by-catch, bullfighting, boiling alive, you name it, and that’s just for starters. In our society, animals still suffer even though we know they feel pain, and we inflict it on them on an industrial scale.

So, then, could the argument be made that we only think generative AI doesn’t have feelings, and that we will see the error of our ways a few decades hence? (Without actually changing our ways then, of course?) Absolutely not—making such an argument would be making category mistakes with reckless abandon. Generative AI models have no interior representations of the world, no sense of self, and they present you not with an answer but a statement about how an answer to your question would probably look like, which is categorically completely different from what an answer actually is. As such, they’re certainly not hurt if you call them names or switch them off. Making such an argument is like saying you should be nice to your car on the off-chance that it turns out to be Christine.

But you should be nice to your car, and to things in general. I’ve always been a great fan of Kondō Marié’s, particularly her philosophy that you should thank an item before discarding it. Why are we repelled by Sid torturing toys in Toy Story? Is it only because toys are sentient in that Pixar universe? Or do we feel that something is terribly off with Sid, and that he won’t stop at torturing toys as a grown-up?

Things (including generative AI), animals, and other human beings have one crucial aspect in common: for every single one of us, objectively, they’re subjectively “not I.” From that perspective, it is plausible that things, just like animals, can become a “training ground” for how we treat other human beings, and that cruelty toward things and animals will sooner or later spill over—call it habituation, virtue ethics, Overton window, whatever you like—to how we treat other human beings. That isn’t to say the only thing that counts at the end of the day is being nice to other human beings. But the effect is observable. There’s a lot of scientific evidence, and numerous papers, on the link between violence toward animals and violence toward humans, and there is the 2009 study “Slaughterhouses and Increased Crime Rates: An Empirical Analysis of the Spillover From ‘The Jungle’ Into the Surrounding Community,” whose title speaks for itself. (“Jungle” refers to the famous Chicago slaughterhouse complex.)

So, yeah. Maybe I read too much Aristotle and think too hard about habit-forming and personality, and maybe I’m too sentimental and nostalgic in general. But I do thank Siri for answering my questions, and I do thank things when I part with them, be it the pair of running shoes that finally fell apart, the terminally worn-out quilt, or the obsolete computer. Am I always nice to people? Well, I try not to be callous, but that doesn’t mean I always know what other people perceive as callous, or rude. Particularly because I’m notorious for being sarcastic by nature and also enjoy a good argumentative fight among friends. However, being callous or rude to things or animals, let alone cruel, would certainly not help me in becoming more considerate of humans.

For all that, I’m bothered by how people will abuse generative AI, especially when taking the tendency into account that many perceive it as a living thing instead of the thing it actually is. I’m bothered by what they are already doing to it, including creating, controlling, and abusing “virtual girlfriends.” I’m bothered by the thought that all this will terribly backfire and spill over and make the world even worse, as bad as it already is, and that in a foreseeable future every year will be like 2014 and 2025 combined.

Addendum
I think SMBC has a point.

Microsoft, via CNET:

The site of the worst nuclear accident in US history could come back into service by 2028 to power Microsoft’s data center energy needs, including artificial intelligence.

According to a purchasing agreement between Microsoft and Constellation Energy, a nuclear reactor at Three Mile Island in Pennsylvania that went offline five years ago would be reactivated. The reactor, dubbed Unit 1, is close to Unit 2, which famously experienced a partial meltdown in 1979 in an incident that cooled enthusiasm for nuclear power in the US for decades.

Pending Nuclear Regulatory Commission approval, the plant would resume operations starting in 2028 and, under a 20-year agreement, would provide energy solely to Microsoft. Constellation said in a press release that it hopes to extend operations there until at least 2054 and that the reactor would generate an additional 800 megawatts of electricity and add 3,400 jobs to the economy.

Oracle, via CNBC:

The electricity demand from artificial intelligence is becoming so “crazy” that Oracle is looking to secure power from next-generation nuclear technology, Ellison told investors on the company’s earnings call Monday.

“Let me say something that’s going to sound really bizarre,” Ellison told analysts. “Well, you’d probably say, well, he says bizarre things all the time, so why is he announcing this one. It must be really bizarre.”

Oracle is designing a data center that will require more than a gigawatt of electricity, the company’s chairman said. The data center would be powered by three small nuclear reactors, he added. “The location and the power place we’ve located, they’ve already got building permits for three nuclear reactors,” Ellison said. “These are the small modular nuclear reactors to power the data center. This is how crazy it’s getting. This is what’s going on.”

Ellison did not disclose the location of the data center or the future reactors.

All this, mind, while even the business rationale for all this is based on bollocks and make-believe.

There’s a distinct type of I believe essentially American megagrift where, to sustain an original fabrication, the lies become bolder and bolder and the audience more gullible and gullible until the bizarreness of it all has reached such titanic proportions that even Tall Tale Land has shrunk to the tiniest speck in the rear mirror—amply exemplified lately by the Orange Fascist, the Nazi Bar Owner, and Snake Oil Sam, who are always eagerly amplified in unchallenged headlines by the breathless stenography of an ad-engagement chasing “CEO Said a Thing” press.

All that, mind again, while there is an exodus of executives at OpenAI and the company is crumbling.

Ed Zitron, pulling no punches:

It’s time to treat OpenAI, Anthropic, Google, Meta, and any other company pushing generative AI with suspicion, to see their acts as an investor-approved act of deception, theft, and destruction. There is no reason to humor their “stages of artificial intelligence”—it’s time to ask them where the actual intelligence is, where the profits are, how we get past the environmental destruction and the fact that they’ve trained on billions of pieces of stolen media, something that every single journalist should consider an insult and a threat. And when they give vague, half-baked answers, the response should be to push harder, to look them in the eye and ask “why can’t you tell me?”

Go and read the whole thing. We’all should be outraged at how these billionaire con artists continue to plunder our cultural resources and poison our information ecosystem while distracting us with this five-stages nonsense along their completely bogus promise of artificial general intelligence as carrot and their equally completely bogus threat of creating a humanity-destroying superintelligence as stick.

As @SwiftonSecurity put it:

Literally every argument about AI risk is entirely made up from exactly nothing. All the terminology is fart-huffing. It has the same evidentiary basis as a story about floopflorps on Gavitron9 outcompeting nuubidons.

And, boy, does Sam Altman have a keen sense of risk and human values! Lobbying for an “oversight” agency—which would effectively create an AI market-controlling monopoly or oligopoly—to prevent humanity-destroying superintelligence from happening has been described, as I wrote about at the time, as calling with a straight face for environmental regulations to prevent the creation of Godzilla. (And don’t miss this Honest Government Ad on the risks of AI!)

Yes, the first cracks do appear (PDF) in the hype machine, and the light you can see falling through these cracks comes from colossal piles of money on fire. Generative AI as such will certainly stay with us and do marvelous things that are both beneficial and evil, but the grandiose earth-shattering, humanity-changing claims from these serial snake-oil merchants will sooner or later go the way of the dodo—just like those grandiose claims around 3D printing, Mars colonies, crypto, blockchain, or Level 5 autonomous cars.

If you’re using grammar and spell checkers, you’re already using LLM-enhanced algorithms whether you want it or not. You can argue whether that counts as “AI use” that needs to be disclosed—I’d say it doesn’t. They’re not doing a better or worse job than dedicated grammar and spell checkers did before.

But let’s look at the next step up: “AI enhanced” spell and grammar checkers that promise to “clarify” your text, make it “more concise,” and similar. Now, the “clear and concise” cudgel has been used—but rarely substantiated beyond the most obviously overcooked examples—since Strunk & White inflicted modern history’s most terrible style guide on English speakers. So I became curious, and I experimented with this “clarify” feature a lot. You really can’t say all it ever does is make your text as bland and uninteresting as (in)humanly possible; it also often suggests “stylish” substitutions with results reminiscent of using more flowery expressions at a dinner conversation after the third martini.

But I actually use that feature from time to time, and not only for its entertainment value!

What I use it for is to check if the original meaning of a complex sentence is retained in the “clarifying” suggestions. If yes, then everything’s fine, nothing needs to be changed and I can move on. If not, then there’s obviously some potential for confusion or misunderstanding there, so I need to go back to my sentence or paragraph to remedy that.

Which is actually not different from the way I work with copyedited manuscripts that came back from the publisher. Whenever the copy editor suggests a rewrite “for clarity,” that rewrite is almost invariably completely wrong. But it’s really helpful, because now I see the potential for confusion and misunderstanding and can rewrite that sentence or paragraph on my own.

Thus, checking whether “AI” correctly understands your complex sentence or paragraph is—I think—a perfectly legitimate use for creative or scholarly writing. Relying on and click-approving such suggestions, however, seems to me neither legitimate nor helpful in these domains.

For other domains, well, go ahead! This “clarify” feature can certainly assist you in squeezing the last drop of personal identity from your business emails.

There are several drafts in my WordPress backend I eventually didn’t bother finishing, partly because I was busy with other things, partly because I feel I’ve wasted enough time and energy writing about AI/LLM/GPT baseline topics here and on my essay on Medium already. But also because I was getting tired of the usual suspects—from antidemocracy-bankrolling techno-authoritarian brunchlord billionaires to hype-cycle-chasing journos who gullibly savor every serial fabulist’s sidefart like a transubstantiated host to singularity acolytes on Xittler who have less understanding of the underlying technologies and attached academic subjects than my late pseudosasa japonica yet corkscrew themselves into grandiose future assertions while flying high from snorting up innuendos of imminent ASI like so many lines of coke.

Thus, I focused on more practical things instead. To start with, I dug into the topic of copyright, with which I’m not yet done. I’m a copyright minimalist and zealous supporter of Creative Commons, Open Science, Library Genesis, Internet Archive, and so on, but you don’t have to be a copyright maximalist to find the wholesale looting of our collective art, science, and culture by that new strain of plantation owners bred from capitalism’s worst actors truly revolting. However, the arguments against the latter have to be very precise against the background of the former, and I’m still at it.

The other thing I did was focusing on aspects—technological, philosophical, social, ethical—that are interesting and a bit more rewarding to think about in the context of large language models. For this roundup, I picked three aspects from the technical side.

The Scaling Process
The first aspect is the scaling process, as observed so far. Making large language models larger not only leads to quantitative improvement but also to qualitative improvement in terms of new abilities like arithmetic or summarization. These abilities are often interpreted as evidence for “emergence,” a difficult (and loaded) term with different definitions in different fields. This recommended article from AssemblyAI from about a year ago is a good introduction, and it also explains why we cannot simply slap the label “emergence” on this scaling process and call it a day. Well, you can define emergent behavior simply as “an ability that is not present in smaller models but present in larger models,” but I believe that’s not only profoundly misguided but actually a cheap parlor trick. While we can’t predict which abilities will pop up at which scale, there’s scant reason to believe that any of them did, do, or will indicate an actual “regime change” that emergence in this definitional context requires, i.e., a fundamental change of the rules that govern a system. But that no “emergence magic” is involved, by all sane accounts, doesn’t make the scaling process any less intriguing.

Interpolation and Extrapolation
About a year ago, there was this fantastic article “ChatGPT Is a Blurry JPEG of the Web” by Ted Chiang in The New Yorker. The article makes several important points, but it might be a mistake to think that large language models merely “interpolate” their training data, comparable to how decompression algorithms interpolate pixels to recreate images or videos from lossy compression. Data points in large language models, particularly words, are represented as vectors in a high-dimensional vector space, which are then processed and manipulated in the network’s modules and submodules. And here’s the gist: for very high-dimensional vector spaces like large language models, there’s evidence that interpolation and extrapolation become equivalent, so that data processing in large language models is much more complex and much more interesting than mere interpolation.

Fractal Boundaries
Finally, a recent paper by former Google Brain and DeepMind researcher Sohl-Dickstein investigates the similarities between fractal generation and neural network training. There’s evidence that the boundary between hyperparameters that lead to successful or unsuccessful training (neural networks) behave like the points from function iteration (fractal generation) that define the boundary where these iterations converge (remain bounded) or diverge (go to infinity). While the generated properties from low-dimensional functions in fractal generation and the complex functions that act in a high-dimensional space in neural network–training differ, these similarities nevertheless might explain the chaotic behavior of hyperparameters in what Sohl-Dickstein calls “meta-loss landscapes” as the space that meta-learning algorithms try to optimize. In a nutshell: meta-loss is at its minimum at the “fractal edge,” the boundary between convergence and divergence, which is exactly the region where balancing is most difficult. Lots of stuff, all very preliminary, but highly captivating.

Originally, I planned this week’s roundup to be specifically about AI/LLM/ChatGPT in research and education, but I pushed these topics back a week in the light of current events. You’ve probably heard by now that Microsoft’s recent Bing AI demo, after some people took a closer look, was a far greater disaster than Google’s Bard AI demo had been a few days earlier.

Dmitri Brereton:

According to this [Bing AI’s] pros and cons list, the “Bissell Pet Hair Eraser Handheld Vacuum” sounds pretty bad. Limited suction power, a short cord, and it’s noisy enough to scare pets? Geez, how is this thing even a best seller?

Oh wait, this is all completely made up information.

It gets worse from there. Predictably, Bing AI also resorts to gaslighting, a topic I touched upon in my recent essay on Artificial Intelligence, ChatGPT, and Transformational Change over at medium.com.

But hey, people want to believe! Which can be quite innocuous, as exemplified by this Golem article in German language. Shorter version: “On the one hand, Bing AI’s answers to our computer hardware questions were riddled with errors, but on the other, it generated a healthy meal plan for the week that we took at face value, so we have to conclude that Google now has a problem.”

Then, it can be stark, ridiculous nonsense. Which especially flowered in Kevin Roose’s “conversational article” in The New York Times, under the original headline “Bing’s A.I. Chat Reveals Its Feelings: ‘I Want to Be Alive.’”

Emily M. Bender gave it the thrashing it deserved:

And then here: “I had a long conversation with the chatbot” frames this as though the chatbot was somehow engaged and interested in “conversing” with Roose so much so that it stuck with him through a long conversation.

It didn’t. It’s a computer program. This is as absurd as saying: “On Tuesday night, my calculator played math games with me for two hours.”

That paragraph gets worse, though. It doesn’t have any desires, secret or otherwise. It doesn’t have thoughts. It doesn’t “identify” as anything. […] And let’s take a moment to observe the irony (?) that the NYTimes, famous for publishing transphobic trash, is happy to talk about how a computer program supposedly “identifies.”

You can learn more about what journalism currently gets terribly wrong from Bender’s essay On NYT Magazine on AI: Resist the Urge to Be Impressed, again over at medium.com. There, she looks into topics like misguided metaphors and framing; misconceptions about language, language acquisition, and reading comprehension; troublesome training data; and how documentation, transparency, and democratic governance fall prey to the sycophantic exultation (my phrasing) of Silicon Valley techbros and their sociopathic enablers (dito).

Finally, there’s the outright pathetic. Jumping into swirling vertiginous abysses of eschatological delusions particularly on Twitter, many seem to believe or pretend to believe that the erratic behavior of Bing’s “Sydney,” which at times even resembled bizarre emotional breakdowns until Microsoft pulled the plug, is evidence for internal experiences and the impending rise of self-aware AI.

Linguist Mark Liberman:

But since the alliance between OpenAI and Microsoft added (a version of) this LLM to (a version of) Bing, people have been encountering weirder issues. As Mark Frauenfelder pointed out a couple of days ago at BoingBoing, “Bing is having bizarre emotional breakdowns and there’s a subreddit with examples.” One question about these interactions is where the training data came from, since such systems just spin out word sequences that their training estimates to be probable.

After some excerpts from OpenAI’s own page on their training model, he concludes:

So an army of low-paid “AI trainers” created training conversations, and also evaluated such conversations comparatively—which apparently generated enough sad stuff to fuel those “bizarre emotional breakdowns.”

A second question is what this all means, in practical terms. Most of us (anyhow me) have seen this stuff as somewhere between pathetic and ridiculous, but C. M. [Corey McMillan] pointed out to me that there might be really bad effects on naive and psychologically vulnerable people.

As evidenced by classical research as well as Pac-Man’s ghosts, humans are more than eager to anthropomorphize robots’ programmed behavioral patterns as “shy,” “curious,” “aggressive,” and so on. That an equivalent to this would be true for programmed communication patterns shouldn’t come as a surprise.

However, for those who join the I Want to Believe train deliberately, it doesn’t seem to have anything to do with a lack of technical knowledge or “intelligence” in general, whatever that is. Not counting those who seize this as a juicy consulting career opportunity, the purported advent of self-aware machines is a dizzyingly large wishful thinking buffet that offers either delicacies or indigestibles for a broad range of sensibilities.

On a final note for today, this doesn’t mean that technical knowledge of how ChatGPT works is purely optional, useless, or snobbish. Adding to the growing number of sources already out there, e.g., Stephen Wolfram last week published a 19,000-word essay on ChatGPT that keeps a reasonable balance between being in-depth and accessible. And even if you don’t agree with his hypotheses or predictions on human and computational language, that’s where all this stuff becomes really interesting. Instead of chasing Pac-Man’s ghosts and seeing faces in toast, we should be thrilled about what we can learn and will learn from LLM research—from move 37 to Sydney and beyond—about decision processes, creativity, language, and other deep aspects of the human condition.

As I wrote more than eight (woah!) years ago in the About section, my secret level/side blog just drafts is one part news ticker with commentary on everything related to games, and one part research-adjacent blog posts about game-based learning and ethics. Discussing current AI models fits that agenda pretty well.

What’s more, I started preparing course materials and practical exercises around AI/LLM/GPT models for the upcoming term in April. These will be balanced topic cocktails for second and sixth term students, revolving around creative assistance (game design, art, coding, and writing), development support (production process and project management), and social ramifications (potentials, risks, economic viability, sustainability, equity/fairness, acceptance, workplace integration, and so on).

Thus, on top of my regular posts, linked list items, or essays, these roundups will serve as food for thought in general, and as a repository for upcoming discussions with my students as well.

Sooooo, let’s go!

1. “Chatbots Could One Day Replace Search Engines. Here’s Why That’s a Terrible Idea.” Will Douglas Heaven in MIT Technology Review, March 29, 2022.

This one’s a bit older. It held up well, but most of the arguments are familiar by now. One aspect, however, is worth exploring. From an interview with Emily M. Bender, one of the coauthors on the paper that led Timnit Gebru to be forced out of Google:

“The Star Trek fantasy—where you have this all-knowing computer that you can ask questions and it just gives you the answer—is not what we can provide and not what we need,” says Bender[.] It isn’t just that today’s technology is not up to the job, she believes. “I think there is something wrong with the vision,” she says. “It is infantilizing to say that the way we get information is to ask an expert and have them just give it to us.”

– – – – –

2. “The Guardian View on ChatGPT Search: Exploiting Wishful Thinking.” The Guardian editorial, Februar 10, 2023.

Since the British Guardian began some time ago to excel in publishing disgustingly transphobic opinion pieces, I all but stopped linking to it. But this one adds an intriguing metaphor to the preceding point of view:

In his 1991 book Consciousness Explained, the cognitive scientist Daniel Dennett describes the juvenile sea squirt, which wanders through the sea looking for a “suitable rock or hunk of coral to … make its home for life.” On finding one, the sea squirt no longer needs its brain and eats it. Humanity is unlikely to adopt such culinary habits but there is a worrying metaphorical parallel. The concern is that in the profit-driven competition to insert artificial intelligence into our daily lives, humans are dumbing themselves down by becoming overly reliant on “intelligent” machines—and eroding the practices on which their comprehension depends.

The operative term here is “practices,” mind. That’s the important thing.

– – – – –

3. “We Asked ChatGPT to Write Performance Reviews and They Are Wildly Sexist (and Racist).” Kieran Snyder in Fast Company, February 2, 2023.

Across the board, the feedback ChatGPT writes for these basic prompts isn’t especially helpful. It uses numerous cliches, it doesn’t include examples, and it isn’t especially actionable. […] Given this, it’s borderline amazing how little it takes for ChatGPT to start baking gendered assumptions into this otherwise highly generic feedback.

[O]ne important difference: feedback written for female employees was simply longer—about 15% longer than feedback written for male employees or feedback written in response to prompts with gender-neutral pronouns. In most cases, the extra words added critical feedback [while the feedback written for a male employee] is unilaterally positive.”

To be expected; sexism and racism and similar nasty stuff is always baked into historical data. To keep any AI trained on historical data from developing racist-aunt-or-uncle opinions, with the potential to ruin a lot more than merely your Thanksgiving family dinner, will keep haunting us as one of the biggest challenges in AI.

If you don’t live under a rock, you might have noticed a few remarkable—but not altogether unpredictable—advances in natural language processing and reinforcement learning, prominently text-to-image models and ChatGPT. And now, every concerned pundit comes out of the woodwork, decrying how terrible these developments are for our educational system and for our starving artists.

Is this deluge of AI-generated images and texts terrible? Of course it is. But don’t let them tell you that this is the problem. It’s only the symptom of a problem.

Let’s start with all those people suddenly feeling deeply concerned about the death of the college essay. Education, if you think of it, should do three things: make children and students curious about as many subjects as possible; give them the tools to develop interests around these subjects; and facilitate the acquisition of skills, knowledge, and understanding along these interests. For these ends, virtually every new technology would be useful one way or another. Our educational systems’ priority, however, is feeding children and students standardizable packets of information—a lot with very short best-before dates stamped on them—for evaluation needs and immediate workplace fitness. Just think of it: the world wide web became accessible for general use around 1994! During all that time, almost thirty years, the bulk of written and oral exams didn’t adapt to integrate the internet, but has been kept isolated from it meticulously. Nor, for that matter, has the underlying system changed of keeping information expensive and disinformation free, an infrastructure into which AI-generated nonsense can be fed now with abandon. And all this gate-keeping for what: when there’s a potential majority in any given country to elect or keep electing fascists into high office, the survival of the college essay probably isn’t the most pressing thing on our plate with regard to education.

Then, the exploitation of artists. Could these fucking techbros have trained their fucking models on work that artists consented to or whose work is in the public domain? That’s what they should’ve done and of course didn’t, but please spare me your pundity tears. While it is thoroughly reprehensible, it’s only possible because at that intersection where the tech and creative industries meet, a towering exploitation machine has stood all along—co-opting or replacing or straightaway ripping off the work of artists and freelance artists, and practically everybody who doesn’t own an army of copyright lawyers, the moment their work becomes even marginally successful.

AI will advance, and everything it can do will be done. Nexus-6 models aren’t right around the corner, but they’re an excellent metaphor. We could try and legislate all those leash-evading applications to death, chasing after them, always a few steps behind and a few years too late, trying to prevent new ways of exploitation while reinforcing old ways of exploitation. Or we could try and change our educational and creative economies in ways that make these applications actually useful and welcome, for educators and artists in particular and humanity and this planet in general.

Today, I made some changes to just drafts.

To begin with, I switched the background color from its original daringfireball-flavored grayish-blue to a more maritime-flavored grayish-green. (It’s called “Dark Slate Gray,” but never mind.) If it’s still grayish-blue, hit CMD-R on Mac or CTRL-R on PC a few times to reload the cached style sheet.

Then, I installed a lightbox plugin and added a “slide” icon to the link class; it will display the post’s Open Graph/Twitter Card image in an overlay if you click on it. (That’s the image you see when the post appears on social media.) I might throw in a few other pictures as well.

Finally, I remembered that I created this side blog back in 2014 not merely for research-adjacent blog posts about game-based learning and game-related ethics; I also created it to link to, and to briefly comment on, external content related to games in general. (Whereby, in linked list–fashion, the post’s headline links to the source, not to the post.) This will recommence soon.

Also, I fixed a few bugs.

That’s it.

Last week, I wrapped up two undergraduate-level game design courses with final workshop sessions, and I wanted them to be an enjoyable experience for everyone.

If there’s one thing teaching should accomplish, it’s this: show students a field, its tools, and its biggest challenges in inspiring ways and help them find and solve challenges which a) they find interesting and b) are manageable at their current level of expertise in terms of skill, knowledge, understanding, and attitude. With that, you provide everything that’s needed: autonomy, mastery, purpose, and also—as solving interesting problems almost always demands collaboration—relatedness.

When I was a student, working together existed mostly as group work or team papers, and I hated both. So I always try to set up my workshop sessions in ways even my freshman self would appreciate. As you might have guessed already, the learning theory I’m trying to follow, to these ends and in general, is constructivism, not cognitivism.

Cognitivism is more or less what you experienced in school. It’s about communicating simplified and standardized knowledge to the students in the most efficient and effective manner, who are supposed to memorize this knowledge in a first step, and then memorize the use of that knowledge in a second step by solving simplified and standardized problems. Sounds familiar, right? Cognitivism is all about memory and mental states and knowledge transfer. It’s not at all about creating curiosity, facilitating exploration, or providing purpose. Cognitivism doesn’t foster and stimulate mastery goals and personal growth (getting really good at something), it focuses instead on performance goals (getting better grades, and getting better grades than others). There’s no autonomy to speak of, only strict curricula and nailed-down syllabi and standardized testing. And there’s no relatedness to speak of either, because there’s no collaboration or cocreation beyond the aforementioned occasional group work or team paper. Because real collaboration and cocreation would mess up standardized testing and individual grading.

Constructivist learning theory, in contrast, is about everything cognitivism ignores: interests, exploration, autonomy, mastery goals, purpose, and relatedness through collaborative and cocreative problem-solving. Certainly, constructivism has its own tough challenges to navigate, but that’s no different from any other learning theory, like cognitivism or behaviorism. According to constructivism, students learn through experiences—along which they create internal representations of the things in the world, and attach meaning to these internal representations. Thus, it’s not about learning facts and procedures but developing concepts and contexts—by exploring a field’s authentic problem space, picking a challenge, and then exploring the field’s current knowledge space and its tool boxes to solve that challenge. The critical role of the teacher is to provide assistance and turn unknown unknowns into known unknowns, so that students can develop a better understanding of the scope, the scale, and the difficulties involved. Without the latter, learning and problem-solving can easily become a frustrating experience instead of an enjoyable one.

And that’s what I mean by enjoyable. In practice, in my final workshop sessions, everybody can pick a problem they’re burning to solve, gather like-minded fellow students, and try to solve that problem together in a breakout room by applying the knowledge and the tools we’ve explored and collected along the course. Or, alternatively, they can stay in the “main track” and tackle a challenge whose rough outlines I set up in advance, moderated by me. In the end, every group presents their findings and shares their most interesting insights.

That’s a setup even my younger self would have appreciated! Everyone can, but by no means has to, shoulder the burden of responsibility by adopting a challenge. Everyone can, but by no means has to, win over other students for a challenge, or join an individual group. Everyone can decide to stay in the main track, which is perfectly fine.

The problem with cognitivism is not that it doesn’t want to die, it’s that is hasn’t noticed it’s already dead. Just think about it: almost everybody can tell, or knows somebody who can tell, an uplifting story about that one outstanding teacher who inspired them despite everything else going down in school. But the very foundation of our educational system is that it’s supposed to work independently of any specific individual, of any specific outstanding teaching performance.

Should we ever design a framework to accomplish this, which might or might not happen in the future, that makes our educational system truly work on the system level, cognitivism certainly isn’t it. Cognitivism is a shambling zombie that, over time, has gradually slipped into its second career as a political fetish.

2020 went by in a daze. For the first time since 2015, I didn’t publish anything, neither a paper nor a book, and I practically stopped blogging. I did write, but I’m months behind my self-set schedule. It wasn’t just the Corona crisis—it were developments in Hong Kong, the U.S., and Israel too that put me into a negative mental state that continually drained my energy.

Now, things have improved in the U.S., but the situation in Hong Kong becomes more terrifying by the minute, and from Israel come the most alarming news. Moreover, I live in Germany—where politicians, propagandists, mercenary scientists, and dangerous alliances of Covid-19 denialists alike torpedo every sensible solution that real scientists and a handful of public figures come up with to fight the virus and keep people safe.

Among the most disastrous measures, in every respect, is the premature opening of schools. This has no rational explanation. Schools in Germany are notoriously underfunded and have no digital infrastructure to speak of. Teachers are bogged down by administrative work. The integration of technologies in the classroom very rarely exceeds the use of calculators and overhead projectors. And, countless political statements of intent notwithstanding, no one ever really gave a shit, and nothing ever changed.

Now, during times of plague, education is suddenly the most terribly precious thing, and to send children back into crowded classrooms is more important than all the people this might kill in the process, or damage for life.

Of course, there are economic reasons to get children out of the way as fast as political decorum allows, going hand in hand with the ministerial refusal to make home office work, where possible, mandatory. Dying for the GDP is something we all understand.

But, considering the tremendous scope of suffering inflicted by Covid-19, that’s too rational an explanation for the consistently irrational demeanor and decision-making, where the state-level secretaries of education push toward opening the schools to strengthen the next pandemic wave each time like clockwork. What’s really going on, and that took me forever to understand, is that “school” is not, or no longer, structured like a place of education & learning. Instead, school is structured like a fetish, forever pointing toward education as a lost referent that can no longer be retrieved.

As such, it is hyper-resistant to any kind of change or reform; to scientific and technological progress; to social and psychological and pedagogical insights.

As such, it wastes twelve or thirteen years of everyone’s life on memorizing swiftly perishable facts instead of teaching curiosity; focuses on tests and grades instead of fostering skill, creativity, and understanding; insists on following the curriculum instead of inspiring students with the love, and thirst, for knowledge and for lifelong learning.

As such, it demands that everyone suffer. And now, true to its nature as a fetish, it demands the willingness to sacrifice your loved ones in times of plague as well.

Today, on the last day of Hanukkah, we will do three things: wrap everything up, briefly; talk about the three missing elements period style, game type, and game loop; and draft a task list with items that would be needed to advance our conceptual sketch toward something pitchable, i.e., a game treatment.

Read More »

During the fourth, fifth, and sixth day of Hanukkah, we sketched a number of design decisions for the Interactivity, Plurimediality, and Narrativity territories.

Today, we will engage the fourth and final design territory of the Ludotronics paradigm, Architectonics.

Instead of merely covering “story” or “narrative,” Architectonics denotes the design and arrangement of a game’s dramatic structure in terms of both its narrative structure and its game mechanics and rule dynamics.

Read More »

Yesterday’s design territory, Plurimediality, was about “compelling aesthetics,” seen from the viewpoint of functional aesthetics and working toward a game’s consistent look-and-feel for a holistic user experience.

Narrativity, our design territory for today, is about “emotional appeal,” seen from the viewpoint of narrative qualities or properties that work toward conveying specific meaning in a specific dramatic unit for a memorable gameplay moment.

For the latter, the Ludotronics paradigm works with four content domains: visual, auditory, kinesthetic, and mythological. Obviously, to be able to plan specific narrative qualities or properties for memorable gameplay moments, we need to know a lot more about our purposeful/educational Hanukkah game than we know at the moment. All we can do right now is sketch a few general principles that we can apply in each domain.

Read More »

While yesterday’s Interactivity territory with its game mechanics, rule dynamics, and player interactions has the “just-right amount of challenge” at its motivational core, Plurimediality is associated with “compelling aesthetics.”

Which, in a game, comprises not only graphics, sound, music, writing, voice acting, and the game’s look-and-feel in general, but also usability in its many forms including player controls. That’s because aesthetics and usability are two sides of the same coin, united by function. Accordingly, Plurimediality integrates design thinking from the perspective of functional aesthetics (informed by the Ludology dimension) and the perspective of aesthetic experience (informed by the Cinematology dimension). One can’t be great without the other, and every element in turn must connect with the theme.

To recap our prospective purposeful/educational Hanukkah game’s concept so far, we’re considering a game with intergenerational cooperative gameplay for Jewish and non-Jewish children and Jewish parents/relatives with a core experience in the ludological dimension, the general theme of “hope,” and a strong focus on the motifs “anticipation,” “ambition,” and “trust” in the Interactivity design territory.

So what’s the aesthetic experience of our Hanukkah game going to be? To answer this question, tentatively, we will look at the game’s style and sound.

Read More »

To recapitulate our insights from the first three days of Hanukkah, our purposeful/educational Hanukkah game’s theme will be “hope”; its core will fall into the ludological dimension; it will have a complex primary and secondary target audience of Jewish and non-Jewish children and Jewish parents and relatives; and its USP will be intergenerational cooperative gameplay.

Today, we will explore the first design territory for our Hanukkah game, Interactivity. Within the Ludotronics paradigm, Interactivity is informed by game mechanical aspects, i.e., the mechanics and rules of the game, and ludological aspects, i.e., how players interact with the game and with other players. (The other three design territories besides Interactivity, to be visited in the following days, are Plurimediality, Narrativity, and Architectonics.)

Read More »

From the research we did on the first and the second day of Hanukkah, we collected a number of design parameters for our purposeful/educational Hanukkah game. These are:

• it’s not well suited for a (pseudo-)historical combat or strategy game;
• it must be playable for non-Jewish players without having to “play-act” Jewish rituals;
• “hope” is a suitable theme for design decisions across all four design territories;
• the best fit for the core playing experience is the ludological dimension.

Today, on the third day of Hanukkah, we will explore possible target audiences and the game’s unique selling proposition, or USP.

Read More »

Yesterday, on the first day of Hanukkah, we sounded out the historical context of the Hanukkah story, thereby eliminating a historical combat or strategy game. Now we will turn to the meaning or meanings that Hanukkah has acquired over time, from the Talmud until today, to find our purposeful/educational game’s theme and its core experience in one of the four dimensions, i.e., Game Mechanics, Ludology, Cinematology, or Narratology.

Read More »

Usually, everything starts with a game idea. From there, you can proceed, step for step, maybe along the Ludotronics paradigm, until you have a pitchable concept—no matter whether it’s a publisher pitch, an in-house pitch, or a war cry to assemble a crackerjack team for your indie title. Now what if someone—a customer, a publisher—approaches you not with an idea, but with a topic?

Along the eight days of Hanukkah, of which today is the first, let’s sketch a purposeful/educational game concept about Hanukkah as an exercise.

Read More »

Game-based learning comes in many different shapes and styles that encompass everything from dedicated educational content in serious games to training simulations to communicating knowledge about historical eras or events and questions of ethnicity or gender in AAA action-adventure games. Yet, as always, well-meant isn’t necessarily well-done. In the case of AAA and high-quality independent games, it is often the lack of “native informants” during the development process that turn good intentions into public meltdowns.

“Native informant” is a term from post-colonial discourse, particularly as developed by Gayatri Chakravorty Spivak. It indicates the voice of the “other” which always runs the risk of being overwritten by, or coopted and absorbed into, a dominant public discourse. Even the term itself always runs this risk.

There are three topics of the “other” I’d like to touch upon in three consecutive posts: gender, ethnicity, and madness. For this post on gender, let’s look at some well-known examples.

With a native informant, the original portrayal of Hainly Abrams wouldn’t have failed as abysmally as it did in Mass Effect: Andromeda. BioWare’s reaction was laudable, certainly—they reached out to the transgender community and made the appropriate changes. But why didn’t they reach out in the first place? Who, on the team, was comfortable with writing the “other” without listening to their voices? Another egregious example were RimWorld’s early gender algorithms—where, in contrast, the developers’ reaction was anything but laudable. (I might be going out on a limb here, but I don’t think OkCupid data qualifies as a native informant.)

Calling in native informants as consultants is exceptionally useful in more than one respect.

First, obviously, they prevent your team from making outrageous mistakes. In the case of Mass Effect: Andromeda, it was a particular mistake that could be fixed with a patch. But when a character’s actions or even whole chunks of the plot are based on faulty premises, that’s not easily fixed at all. Enter Assassin’s Creed: Odyssey’s DLC “Legacy of the First Blade: Shadow Heritage,” where Cassandra was “hetconned,” i.e., forced into a heterosexual relationship to become a parent.

Then, why would you want to stop at merely not getting it wrong? Native informants will provide you with involving details that will turn cliche attitudes into motivated actions & emotions and transform your cardboard characters into engaging personalities.

Finally, hiring native informants as consultants adds diversity to your team, enriches your game, and broadens your reach beyond your mainstream audience. Toward people, that is, who would gladly buy your game if they see themselves represented in it—in the contexts of visibility, of acceptance, and of role models as drivers of empowerment.

Among the great advantages of game-based learning is, in well-designed games, that players are “tested” exclusively on skills, knowledge, understanding, or attitude that they have learned, or should have learned, while playing the game. Certainly, there are exceptions. Games can be badly designed. Or, problems of familiarity might arise, as discussed in Ludotronics, when conventions of a certain class of games (misleadingly called “genre”) raise the Skill Threshold for players who are not familiar with them and might also interfere later in test situations. Most of the time, though, players are indeed tested on what the game has taught them, and well-designed “term finals” like level bosses do not simply test the player’s proficiency, but push them even further.

This rhymes beautifully with the rule “you should not grade what you don’t teach” and its extension “if you do teach it, grade it only after you taught it.” While this sounds like a no-brainer, there are areas in general education that have a strong tendency toward breaking this rule, with grammar as a well-known example. And then there’s the problem of bad or insufficient teaching. When students, through no fault of their own, failed to acquire a critical skill or piece of knowledge demanded by the advanced task at hand, how would you grade them? Is it okay, metaphorically speaking, to send students on bicycles to a car race and then punish them for poor lap time performances? But if you don’t grade them on the basis of what they should have learned, doesn’t that mean lowering the standards? As for the students, they can’t just up and leave badly designed education like they can stop playing a badly designed game.

Another aspect, already mentioned, is that tests in games not only test what has been taught, but are designed to push player proficiency even further. This is possible, and possible only, because the player is allowed to fail, and generally also allowed to fail as often as necessary. Such tests, moreover, can become very sophisticated, very well-designed. In Sekiro: Shadows Die Twice, you might be able to take down a boss with what you’ve already learned instead of what you’re supposed to be learning while fighting that boss, but you’ll be punished for it later when you desperately need the skill you had been invited to learn. You can read all about this example in Patrick Klepek’s “15 Hours In, Sekiro Gave Me a Midterm Exam That Exposed My Whole Ass,” which is recommended reading.

The second reverse question I love to discuss in my lecture on media psychology/media didactics for game designers is to ask, instead of what games can bring to the classroom, what the classroom brings to the table compared to games. Again, it’s about non-specialized public gymnasien in Germany, so the results are not expected to be particularly invigorating.

What about motor skills, dexterity, agility, hand-eye coordination, reaction times, and so on? What about endurance, persistence, ambition (not in terms of grades), or patience? Here, the classroom has little to offer. Possible contributions are expected to come from physical education, certainly. But—with the exception of specialized boarding schools, or sportgymnasien—physical education is almost universally despised by students for a whole raft of reasons, and it is not renowned for advancing any of the qualities enumerated above in systematic ways.

What about »Kombinationsfähigkeit« in terms of logical thinking, deduction, and reasoning? What about strategy, tactics, anticipatory thinking (which actions will trigger which events), algorithmic thinking (what will happen in what order or sequence), and similar? Again, the classroom in non-vocational or non-specialized schools has little to offer here, if anything at all.

Finally, what about creativity, ingenuity, resourcefulness, imagination, improvisation, and similar? Some, the lucky ones, have fond memories of art lessons that fostered creativity. But even then: on this palette, creativity is just one color among others. Music lessons have potential too, but—barring specialized schools again—it’s rare to hear students reminisce lovingly about music lessons that systematically fostered any of these qualities.

Now, the curious thing is that serious games and game-based learning projects have a tendency to try and compete precisely with what the classroom does bring to the table for the cognitive domain, notably in its traditional knowledge silos that we call subjects. This, not to mince words, is fairly useless. Serious games, in careful studies against control groups, almost never beat the classroom significantly in terms of learning events, learning time, depths of knowledge, or even knowledge retention—but they come with a long time to market, a stiff price tag, and tend to burn content like a wildfire. Instead, game-based learning should focus on the psychomotor domain, the affective domain, those parts of the cognitive domain that the classroom notoriously neglects, and rigorously unsiloed knowledge from every domain. And if all that can’t be integrated into the classroom because the classroom can’t change or adapt and integrate games and new technologies in general, then let’s build our own classroom, maybe as a massively multilearner online game-based learning environment. The century’s still young.

In my lecture on media psychology/media didactics for game designers, reverse-questioning how school lessons and school curricula relate to technology and games counts among my favorite exercises. For example, instead of asking what technology in general and games in particular can do for the classroom, we ask which technological advances schools have actually adopted since the middle of the twentieth century.

Now, as I am teaching in Germany, and most of my students come from non-specialized public gymnasien, you can probably guess where this is going. With the rare exception of interactive whiteboard use (roughly one student out of fifteen), the two technological advances properly integrated into regular classroom use in seven decades are [check notes] the pocket calculator and [check notes again] the overhead projector.

Historically, all technological advances that have been proposed for classroom use in Germany, including pocket calculators and TV/VCR sets and computers and cell phones and tablets, and so on, were viciously opposed by teachers, parents, administrators, and politicians alike with an inexhaustible reservoir of claims ranging from the decline of educational integrity (»Kulturverfall«) to wi-fi radiation panic (»Strahlenesoterik«). Today, the copy & paste function especially is viewed as a creeping horror that must never be allowed to make it past the hallowed doorstep of higher education.

Add to that an often desolate financial situation. There are cases where schools can’t afford a decent wi-fi infrastructure, desktop or mobile hardware, software licenses, or, especially, teacher training. But all these problems could be surmounted in principle if it wasn’t for the one titanic challenge: the curriculum. Cognitivism has brought us this far, and we certainly shouldn’t abandon it. But it should become part of something new, a curriculum based on what and how we should teach and what and how we should learn in the 21st century. This requires motivational models that are up to the task; we can start with Connectivism and with models like flow and self-determination, both of which feed the Ludotronics motivational model, and advance from there. Technological, social, and other kinds of advances should become deeply integrated into this new curriculum. To turn our information society into a true knowledge society, we must leave the era of age cohorts, classes, repeated years, and silo teaching behind and embrace learning and teaching as a thrilling and, before all else, shared quest of lifelong exploration.

Recently, I asked about the possible effects of “one-size-fits-all educational methodology with predetermined curricula and standardized testing” and, especially, “conditioned learning of siloed educational subjects detached from personal experience in large classes solely determined by year of birth.” That, of course, was purely a rhetorical question as these effects are clearly visible to everybody. Rarely do schools instill in us a deep and abiding love for learning and for the subjects taught, the historical events envisioned, the works of literature read, the math problems solved. Rarely do we fondly remember our school days and sentimentalize them into nostalgic yearnings for a lost pleasure. In my 2013 inaugural lecture “Knowledge Attacks!: Storyfication, Gamification, and Futurification for Learning Experiences and Experiential Learning in the 21st Century” which, alas, I still haven’t managed to put online, I developed two sample scenarios to break down educational silos in our schools.

The first sample scenario took off from the topic of “calculus,” typically siloed and restricted to “math” courses. Why don’t we confront students instead with the exact same problems Newton and Leibniz faced and connect the invention of calculus with learning content from its historical context: Newton’s Principia and the Great Plague of London; the English civil war and the Bill of Rights; the baby and toddler years of the Scientific Method and the principles of rational inquiry; the Great Fire of London and modern city planning; the formation of the United Kingdom and colonial power struggles; the journalist, author, and spy Daniel Defoe and Protestant work ethic; the transition from Baroque to Rococo in painting and music; the Newtonian telescope and Newton’s laws of motion; the dissociation of physics and metaphysics. Interim summary: math, physics, philosophy, religion, English language, history, geography, music, art, literature, astronomy. Oh, and sports: the evolution of cricket during this time in post-restoration England as the origin of professional team sport.

Or “democracy,” aspects of which are usually siloed in history courses and/or a variety of elective courses like “politics” or “citizenship.” What if students were given the task, perhaps in a time-travel setup, to convince the Athenian assembly in 482 B.C.E. via political maneuvering to distribute the new-found silver seam’s wealth among the Athenian citizens instead of following Themistocles’s proposal to build a naval fleet? So that the Battle of Salamis would never happen, the Greek city states swallowed by the Persian empire, neither the Roman republic nor large parts of the world become hellenized, and the European Renaissance as we know it would never happen? The potential in terms of learning content: the dynamics of democracy and political maneuvering; the history of classical antiquity; general economics, trade, and the economics of coins and currencies; the structure and ramifications of democratic systems built on slave economies; Greek language; rhetoric; comedy and tragedy; myth; Herodotus and the beginnings and nature of historical writing; geometry; the turn from pre-Socratic to Socratic philosophy; sculpture and architecture; ship building; astronomy; geography; the Olympic Games. Strategy, probably—especially naval strategy if the plan to change history fails to succeed: Salamis from the point of view of Themistocles on the side of the Greeks and from the point of view of Artemisia—the skillful and clear-sighted commander of a contingent of allied forces in Xerxes’s fleet—on the side of the Persians. Infinite possibilities.

Now, what’s being introduced in Finland right now as “cross-subject topics” and “phenomenon-based teaching” looks quite similar in principle to my developing concept of “scenario learning.” As the Independent's headline puts it, “Subjects Scrapped and Replaced with ‘Topics’ as Country Reforms Its Education System”:

Finland is about to embark on one of the most radical education reform programmes ever undertaken by a nation state—scrapping traditional “teaching by subject” in favour of “teaching by topic.” […] Subject-specific lessons—an hour of history in the morning, an hour of geography in the afternoon—are already being phased out for 16-year-olds in the city’s upper schools. They are being replaced by what the Finns call “phenomenon” teaching—or teaching by topic. For instance, a teenager studying a vocational course might take “cafeteria services” lessons, which would include elements of maths, languages (to help serve foreign customers), writing skills and communication skills.

More academic pupils would be taught cross-subject topics such as the European Union—which would merge elements of economics, history (of the countries involved), languages and geography.

This sounds exciting already, but there’s more:

There are other changes too, not least to the traditional format that sees rows of pupils sitting passively in front of their teacher, listening to lessons or waiting to be questioned. Instead there will be a more collaborative approach, with pupils working in smaller groups to solve problems while improving their communication skills.

Many teachers, of course, aren’t exactly thrilled. But a “co-teaching” approach to lesson planning with input from more than one subject specialist has also been introduced; participating teachers receive a “small top-up in salary,” and, most importantly, “about 70 per cent of the city’s high school teachers have now been trained in adopting the new approach.”

And even that’s not the end of it. A game-based learning approach, upon which my scenario learning concept is largely built, will also be introduced to Finland’s schools (emphases mine):

Meanwhile, the pre-school sector is also embracing change through an innovative project, the Playful Learning Centre, which is engaged in discussions with the computer games industry about how it could help introduce a more “playful” learning approach to younger children.

“We would like to make Finland the leading country in terms of playful solutions to children’s learning,” said Olavi Mentanen, director of the PLC project.

Finally, from the case studies:

We come across children playing chess in a corridor and a game being played whereby children rush around the corridors collecting information about different parts of Africa. Ms. Jaatinen describes what is going on as “joyful learning.” She wants more collaboration and communication between pupils to allow them to develop their creative thinking skills.

What I feel now is a powerful urge to immediately move to Finland.

(h/t @JohnDanaher)

The first #ECGBL2014 presentation I attended was “Experimenting on How to Create a Sustainable Gamified Learning Design That Supports Adult Students When Learning Through Designing Learning Games” (Source) by Charlotte Lærke Weitze, PhD Fellow, Department of Education, Learning and Philosophy, Aalborg University Copenhagen, Denmark.

Weitze’s paper relates to a double challenge so crucial for designing game-based learning solutions that you’ll see me coming back to it on this blog time and again. One side of this challenge is students burning through content way faster than educators and game designers are able to produce; you just can’t keep up. The other side is the challenge of practicing complex, non-siloed learning content within game-based learning environments that are incompatible with practices such as rote learning, flash cards, repeat training, or standardized testing.

In other words, it’s non-educational games’ “replay value” challenge on screaming steroids. Among the most promising solutions is mapping the “learning by teaching” approach to designing content, i. e., having students design fresh game content as a learning-by-teaching exercise. Obviously, both effectiveness and efficiency of this approach depend on the number of students that are part of the system, where output will exponentially take off only after a certain threshold has been reached, both numerically and by way of network effects.

Of course there are pitfalls. One obvious pitfall is that students—with the exception of game design students, of course—are students, not game designers. And while, along more reasonable approaches, they don’t have to design fully functional games by themselves with game mechanics, rules, and everything, it’s still hard to design entertaining, i. e., “playable” content for games without at least some game design knowledge from various fields—prominently balancing, guidance strategies for player actions, or interactive storytelling. This was one of the reasons why, regrettably, Weitze’s experiment, where students also had to build the content with software, fell short.

As for the experimental setup (which employs the terminology of James Paul Gee’s more interaction-/discourse-oriented differentiation between little “g” games and big “G” Games), students design and build little “g” games for a digital platform within the framework of a “gamified learning experience,” the big “G” Game, which includes embedding learning goals and evaluating learning success. The little “g” games in this case are built for other students around cross-disciplinary learning content from the fields of history, religion, and social studies in a three-step process consisting of “concept development, introduction and experiments with the digital game design software (GameSalad), and digital game design” (4). Moreover, this whole development process (within the big “G” Game) had to be designed in such a way as to be motivating and engaging for the students on the one hand, and to yield evaluable data as to its motivational impact on individual learning successes on the other.

Experiences from this experiment, unsurprisingly, were described in the presentation as “mixed”—which is academic parlance for “did not fucking work as planned at all.” The problems with this setup are of course manifold and among these, “focus” is the elephant in the room.

There are far too many layers, especially for an experiment: the gamification of game design processes for learning purposes (the big “G” Game) and the design thereof with in-built learning goals and evaluation strategies; below that, then, the game design processes for the little “g” games including, again, learning goals and evaluation strategies. In other words, the students were supposed to be learning inside the big “G” Game by designing little “g” games in groups with which other students in turn would be able to “learn from playing the games and thus gain knowledge, skills and competence while playing” (3)—all that for cross-disciplinary content and executed in a competitive manner (the big “G” Game) by 17 students and three teachers none of which had sufficient game design experience to start with, and under severe time constraints of three workshop sessions of four hours each.

Besides “focus,” the second elephant in the room is “ambition,” which the paper acknowledges as such:

In the current experiment the overall game continued over the course of three four‐hour‐long workshops. Though this was a long time to spend on an experiment, curriculum‐wise, for the upper‐secondary students, it is very little time when both teachers and students are novices in game‐design. (3)

Plus:

This is an ambitious goal, since a good learning‐game‐play is difficult to achieve even for trained learning game designers and instructors. (3)

And the nail in the coffin, again unsurprisingly:

At this point [the second workshop] they were asked to start considering how to create their game concept in a digital version. These tasks were overwhelming and off‐putting for some of the students to such a degree that they almost refused to continue. This was a big change in their motivation to continue in the big Game and thereby the students learning process was hindered as well. (7)

Finally, what also generated palpable problems during this experiment was the competitive nature of the big “G” Game. While the paper goes to some lengths to defend this setup (competition between groups vs. collaboration within groups), I don’t find this approach to game-based learning convincing. Indeed I think that for game-based learning this approach—competition on the macro-level, collaboration on the micro- or group-level—has it upside down: that’s what we already have, everywhere. Progress, in contrast, would be to collaborate on the macro-level with stimulating, non-vital competitive elements on the micro- or group level.

Game-based learning can provide us the tools to learn and create collaboratively, and to teach us to learn and create collaboratively, for sustained lifelong learning-experiences. Competition can and should be involved, but as a stimulating part for a much greater experience, not the other way round. The other way round—where collaboration has to give way to competition as soon as things threaten to become important—is exactly what game-based learning has the potential to overturn and transcend in the long run.

Last year’s ECGBL 2014 (8th European Conference for Game-Based Learning), October 9–10, in Berlin, which included the satellite conference SGDA 2014 (5th International Conference on Serious Games Development & Applications), was located at Campus Wilhelminenhof in Berlin Oberschöneweide and hosted by the University of Applied Sciences for Engineering and Economics HTW Berlin. The conference was opened by Dr.-Ing. Carsten Busch, program chair and professor of media economics/media information technology, followed by an introduction to the HTW Berlin by Dr.-Ing. Helen Leemhuis, faculty dean and professor of engineering management.

Then, the keynote. Oh well.

How to put this politely. To be sure, there are occasions and circumstances where it is a good idea to engage with stakeholders outside academia by inviting industry representatives to academic conferences as keynote speakers, but in this case it rather wasn’t.

The invited speaker was Dr. jur. Maximilian Schenk, formerly “Director Operations and member of the management team” of the German VZ network (which has gone down in history for, among other things, setting the bar for future copy-&-paste operations spectacularly high by copying Facebook literally wholesale, down to style sheets and code files named fbook.css or poke.php), at present managing director of the BIU (Bundesverband interaktive Unterhaltungssoftware / German Trade Association of Interactive Entertainment Software). He addressed his audience of highly qualified postgraduate, postdoc, and tenured veteran researchers from the fields of game-based learning and serious games across a wide range of disciplines verbatim with:

You are the specialists so I won’t go into your terrain, so instead I will tell you something about the fundamentals of serious games that you have to understand to know what making serious games is all about.

During the stupefied silence that followed, Maximilian Schenk acquainted the audience with the BIU and its sundry activities, explained how the traditionally bad image of gaming in Germany, including serious games, was changing as it had been found out that “games make people smarter” (evidence: Spiegel), pontificated about “games as a medium” from “tent fires, maybe 10,000 years from now” to today’s video games, and ended his thankfully brief keynote with an enthusiastic barrage of growth forecasts relating to game-based learning/serious games industries whose outrageously optimistic numbers were inversely proportional to the amount of actual evidence corroborating these numbers.

While appreciated in general, it was rather obvious that the keynote’s briefness had taken the organizers by surprise, and Carsten Busch jumped in to introduce, with all the little tell-tale signs of hurried improvisation, the Swedish Condom08 gamification project. This presentation—its general drift into inappropriately didactic terrain notwithstanding (“What did you learn?”; “What technologies were used?”)—turned out to be enjoyable and stimulating.

And so the conference began. Follow-up posts are in the pipeline.

This week, I’m off to the ECGBL 2014 in Berlin, including the SGDA 2014 satellite conference.

Not sure if I will have time for blogging but if not, bear with me—there’ll be no shortage of posts about conference papers soon.

If anyone reading this blog also happens to attend this year’s ECGBL and wants to connect, check betweendrafts.com/hello-world for every possible channel to get in touch.

The negative motivational potential of programming textbooks and tutorials is second only to the motivational potential of how we teach math. And while visual “learning by doing” systems like Khan Academy’s Computer Programming online course seem like progress, they’re sugarcoating the problem instead of providing a solution.

Bret Victor, two years ago, in his eye-opening essay “Learnable Programming”:

We often think of a programming environment or language in terms of its features—this one “has code folding”, that one “has type inference”. This is like thinking about a book in terms of its words—this book has a “fortuitous”, that one has a “munificent”. What matters is not individual words, but how the words together convey a message.

Everything else—a must-read—follows from there toward building a mental model. The concept of building a mental model is based on an interesting premise: “A programming system has two parts. The programming ‘environment’ is the part that’s installed on the computer. The programming “language” is the part that’s installed in the programmer’s head.” The inspiration to build on this premise, as Victor remarks, came from Will Wright’s thoughts on interactive design in “Sims, BattleBots, Cellular Automata God and Go: A Conversation with Will Wright” by Celia Pearce:

So what we’re trying to as designers is build up these mental models in the player. The computer is just an incremental step, an intermediate model to the model in the player’s head. The player has to be able to bootstrap themselves into understanding that model. You’ve got this elaborate system with thousands of variables, and you can’t just dump it on the user or else they’re totally lost. So we usually try to think in terms of, what’s a simpler metaphor that somebody can approach this with? What’s the simplest mental model that you can walk up to one of these games and start playing it, and at least understand the basics? Now it might be the wrong model, but it still has to bootstrap into your learning process. So for most of our games, there’s some overt metaphor that allows you approach the simulation. (Game Studies: The International Journal of Computer Game Research Vol.2 Issue 1, July 2002)

That, of course, holds important implications for game-based learning design—not just for teaching programming, but that’s a particularly obvious example. In game-based learning design, bootstrapping must take place both at the system level and the content level: the GBL model must be designed in such a way that you “can walk up to it and start playing,” and the same thing applies to how the learning content is designed so that the player ”can walk up to it and start learning.” (A fine example of Hayden White’s The Content of the Form principle at work, incidentally.)

As of now, there is no shortage of games that try to teach programming to kids, but just browsing the blurbs opens jar after jar crawling with inadequacies, to put it mildly. Games aimed at teens tend to float belly up the next time you check (yes I’m looking at you, CodeHero), or remain perpetually promising but unfinished like Dre’s Detox. And you won’t find anything remotely suitable for adults or seniors.

Obviously, if we managed to create games that teach how to code along the lines imagined by Bret Victor, we’d create new generations of coders who could put to good use not only the principles they’ve learned, but the principles with which they’ve learned what they’ve learned, to create great game-based learning designs and experiences for the future.