Those of us who design learning share assumptions we rarely test: that more data on learners improves teaching, that finishing a course means learning from it, and that richer media teaches better. On launch day, millions of adults spent a full work day playing Grand Theft Auto VI (GTA VI). The game deliberately leaves out data, most players never finish it, and progress comes from peer guidance shared as text in Discord channels and social media. Read against decades of learning science, the game puts each of those assumptions to the test. What we stand to learn is instructive if not flattering, for the game as much as for us.
On Thursday, 19 November 2026, Grand Theft Auto VI went on sale after a 13-year wait, and a very large number of people sat down to be taught by it. Take-Two Interactive, its publisher, estimates that the game took in $3.79 billion in its first day, with 32.5 million copies sold before release. An independent analyst, Sensor Tower, puts the early figures lower.
A smaller number stays with me. The research firm Circana reports that PlayStation 5 players averaged 513 minutes in the game on release day. That adds up to 8.5 hours on a Thursday.
Those of us who design learning for adults are told, constantly, that adults have no time and less attention, so we keep shortening courses and end them with a quiz. Say hello to “microlearning” and “one-minute learning”. (Carried to its logical conclusion, “zero-learning” is the ultimate learning outcome…)
Yet here were adults in their millions giving a working day to the rules of an unfamiliar city, from how its police think to how to prepare a robbery that will not fall apart. They did so without a calendar invitation or the promise of a certificate. Victoria Marsick and Karen E. Watkins have spent decades describing this kind of learning, which happens outside formal structures, wherever people have a need and an opportunity to learn (Marsick & Watkins, 2001). Most of what adults know about their work, they learn this way.
So, what is this game teaching, and how?
Gamification copies the cheapest parts of games, so I am taking a second look at GTA VI
In 2018 I wrote that gamification was a disaster for humanitarian learning (Sadki, 2018). I described the fashion for points, badges, and leaderboards as “lipstick on the ugly pig of behaviorism”. I was younger, and I enjoyed the phrase more than I should have.
I would still make the argument underneath it. A game good enough to teach anything costs a fortune, so organizations with small budgets bolt a few rewards onto a dull course and call the result a game. The evidence against those rewards is older than the fashion. A meta-analysis of 128 experiments found that expected and tangible rewards undermine people’s intrinsic motivation, including rewards given simply for taking part or for finishing a task (Deci et al., 1999). Most video games are also closed systems that cannot model the tangle of real work. And the culture around games mirrors societal racism, sexism, and violence.
I have not changed my mind about badges (Sadki, 2012). But a critic owes his subject a second look, and Grand Theft Auto VI is the occasion for one, because gamification is a cheap imitation of what this game does. James Paul Gee made the case two decades ago that good games survive only if players can learn them, and that designers could have solved their learning problem by making games shorter and easier, which players would not have bought (Gee, 2003). His point applies directly to corporate learning that keeps making courses shorter and easier in the hope that more people will finish them.
Education gathers ever more data to control learners, but GTA VI shows that a world which knows less can teach more
The IGN review notes that officers in the game can act only on the information they have, and that the player has no radar showing where they are. If nobody saw you, nobody is looking for you, and if somebody did, the consequences follow from that fact alone. In life, too, what happens next depends on who noticed and what they decided to do about it.
The typical workplace course works the other way. An omniscient system records every click and awards a score that corresponds to nothing outside itself. The learner soon works out that the task is to satisfy the scorekeeper, and does so. The social scientist Donald Campbell described the mechanism in 1979: “The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor” (Campbell, 1979).
Behind that system sits a belief common in educational technology: the more we know about what learners do, the better we can teach them. I hold part of that belief myself. At the foundation, we read learner signals while a course runs and change the design in response (Sadki, 2024d). Bill Cope and Mary Kalantzis, who have thought harder about this than most, describe big data in education as carrying both the promise of an information-rich future and the threat of surveillance (Cope & Kalantzis, 2016). They separate data that learners generate by doing meaningful work from data collected incidentally as they move through a platform. A log of clicks belongs to the second kind. It mostly records whether the learner pleased the system.
The field has tended to answer that problem with more data. George Siemens, who helped establish learning analytics as a discipline, argued in 2013 that it needed to capture far more data so that analysis could reflect the complexity of learning (Siemens, 2013). A decade later, reviewing the research, Oleksandra Poquet found that studies often treat noticing what a dashboard shows as if it were understanding what it means (Poquet, 2024), a confusion I wrote about in 2024 (Sadki, 2024c).
Data helps when it illuminates what learners are doing. Used to control them, it teaches them to perform for the system, and training departments keep paying to install the scorekeeper that Rockstar removed.
Leonida’s police force gives the player something better than a score. John Hattie and Helen Timperley’s review of feedback research found that feedback works when it answers three questions: where am I going, how am I going, and where to next (Hattie & Timperley, 2007). Praise, they found, is rarely effective because it carries so little information about the task.
A witness who calls the police tells the player at once how things are going and what to do next. TechRadar reports similar restraint elsewhere in the game. Players can talk their way out of a confrontation, or disable an opponent without killing, and exercise, food, and sleep govern the character’s health and stamina. The game builds a world in which rest matters, but players first need to figure out the ‘why’ and the ‘how’.
We forgive players who never finish GTA VI but call learners who leave a course failures
Most people who buy this game will never finish it. Trophy data compiled by the Swiss retailer Digitec last year showed that 16.8% of PlayStation owners of Grand Theft Auto V had completed its story. For Red Dead Redemption 2, from the same studio, the figure was 25.1%. These numbers count everyone who owns the game, including people who barely started. The new story runs to 91 missions, and reviewers took between 42 and 115 hours to reach the end.
When the researcher Katy Jordan analyzed 221 massive open online courses, she found a median completion rate of 12.6% (Jordan, 2015).
For a decade, low completion rate has been offered as proof that online education failed.
Nobody says that Grand Theft Auto V failed.
We understand, without being told, that the player came for something, found it, and left satisfied, whether that was a weekend of mayhem or a drive along the coast with the radio on. We do not demand that the player sit through the credits to prove the money was well spent.
I have argued for years that learners deserve the same reading (Sadki, 2025b): a nurse who takes what she needs from a course and goes back to her patients has not dropped out of anything.
GTA VI proves that plain text written by peers spreads knowledge faster than spectacle or cascade training
Before the game had unlocked, written guides to its 55 trophies were already online, followed within hours by checklists and walkthroughs. One estimate puts full completion at 200 hours, and people intend to help one another through all of them. When a player gets stuck, one of the most lavish audiovisual productions ever made sits glowing on the screen while the player reaches for a telephone to read plain text, written by a stranger, for free.
I have made the case for text before, sometimes to polite disbelief (Sadki, 2025c). Text is cheap and quiet, and it can be searched and corrected. A video or a simulation is content, and the learning is whatever the learner then does with it (Sadki, 2025a). Players act on this every time they open a guide.
The stranger who wrote the guide is doing something Lev Vygotsky described almost a century ago. He defined the zone of proximal development as the distance between what a learner can do through independent problem solving and what the same learner can do “under adult guidance or in collaboration with more capable peers” (Vygotsky, 1978, p. 86). The more capable peer, in this case, is whoever solved the problem last night.
Constance Steinkuehler and Sean Duncan found the same pattern when they studied discussion forums for World of Warcraft. In their sample, 86% of the posts were players building knowledge together, not social banter, and 65% treated knowledge as an open process of evaluation and argument (Steinkuehler & Duncan, 2008).
Players also wrote those guides. Much of the world still trains its workforce through a cascade, in which a few people are trained to train others in turn, and that model loses something at every step (Sadki, 2024e). Knowledge about this game travels sideways instead.
George Siemens called this connectivism in 2005. In his account, learning “can reside outside of ourselves”, and “the connections that enable us to learn more are more important than our current state of knowing” (Siemens, 2005).
At the peak of launch, the analytics firm Streams Charts counted 7.91 million people watching others play. Jean Lave and Etienne Wenger described how newcomers learn by taking part at the edge of a community of practice and moving, over time, toward full participation (Lave & Wenger, 1991). Some of those viewers will be posting their own guides by spring.
I have spent years building the same arrangement for a different public. At The Geneva Learning Foundation, health workers in 137 countries write about a real problem in their own district, then give feedback to three colleagues and receive it from three, guided by a shared rubric (Sadki, 2024b). It scales in a way that expert coaching cannot (Sadki, 2024a).

Players built the same arrangement for themselves, and that should settle whether adults will teach each other, unpaid, when the problem matters to them.
Games cost too much to build and forgive too much to prepare anyone for real work
In 2018, I cited $50 million and $200 million as the price of a ‘serious game’. No ministry, school, or employer will ever build a world like this one. On a training budget, what gets built is the points and badges I criticized then.
The subject matter gives me pause as well. The game is rated for mature audiences, and its heroes rob banks. Reviewers describe more room for restraint than the series has offered before. Two days are not enough for me to know whether this game deserves the charge I levelled at game culture in 2018.
Grand Theft Auto VI also launched as a single-player experience, with no online mode. Fail a mission, and a checkpoint returns you to the moment before the mistake. That forgiveness is much of the pleasure.
Failure can teach a great deal.
Manu Kapur found that students who struggled, and failed, with complex problems without any support later outperformed peers who had solved well-structured problems, because working through the next problems let them see what their failed attempts had missed (Kapur, 2008). The failure taught because the students carried it forward. A checkpoint erases it.
That is why the learning stays inside the game. Whether skills learned in one setting carry over to a distant one has been argued for a century without resolution.
Susan Barnett and Stephen Ceci traced the stalemate to studies that fail to specify how far apart the two settings are, and they proposed nine dimensions on which that distance can be measured (Barnett & Ceci, 2002).
On almost any of them, Vice City is far from the work that matters, which happens in clinics, classrooms, small businesses, and families, where nothing resets and other people live with what you did.
Good learning keeps GTA VI’s trust in people and judges itself by what changes in the world
I would not copy this game, even if I could afford to. Rather, I would take its trust in people, who are put in front of a real problem and left to work without an hour of preliminaries.
That trust has critics that deserve an answer. Paul Kirschner, John Sweller, and Richard Clark argued that minimal guidance fails novices, who need strong, direct instruction (Kirschner et al., 2006). They are right that people left alone with a hard problem often flounder.
But the players of Grand Theft Auto VI are not left alone. They have guides, walkthroughs, and millions of more capable peers.
The guidance has moved from the instructor to the network, and it can be found when the learner needs it, at the point of need.
So I would replace the scoreboard with consequences that arise from the work itself, and with what Cope and Kalantzis call recursive feedback, which turns assessment into a continuing dialogue (Cope & Kalantzis, 2017).
I would let peers and plain words carry knowledge sideways, and treat learners as people who design their own social futures, as the New London Group, which Cope and Kalantzis helped found, proposed thirty years ago (New London Group, 1996).
I would add the one thing no game can offer, which is a world that stays changed. For anyone who teaches adults, I know of only one test that holds. What is different now, and for whom?
Some weeks from now, millions of people will set down the controller for the last time. A few will have finished. Most will not, and they will be right not to mind. Leonida will be exactly as they found it.
References
Barnett, S. M., & Ceci, S. J. (2002). When and where do we apply what we learn? A taxonomy for far transfer. Psychological Bulletin, 128(4), 612–637. https://doi.org/10.1037/0033-2909.128.4.612
Campbell, D. T. (1979). Assessing the impact of planned social change. Evaluation and Program Planning, 2(1), 67–90. https://doi.org/10.1016/0149-7189(79)90048-X
CNBC. (2026, November 20). Take-Two shatters sales record as “GTA VI” launches to massive fanfare. CNBC. https://www.cnbc.com/amp/2026/11/20/take-two-sales-as-gta-vi-launches-to-massive-fanfare.html
Cope, B., & Kalantzis, M. (2016). Big data comes to school: Implications for learning, assessment, and research. AERA Open, 2(2), 1–19. https://doi.org/10.1177/2332858416641907
Cope, B., & Kalantzis, M. (Eds.). (2017). e-Learning ecologies: Principles for new learning and assessment. Routledge. https://doi.org/10.4324/9781315699935
Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 125(6), 627–668. https://doi.org/10.1037/0033-2909.125.6.627
Digitec Galaxus. (2025, March 11). Hardly anyone finishes games: So why are they so long? Digitec. https://www.digitec.ch/en/page/hardly-anyone-finishes-games-so-why-are-they-so-long-37002
Forbes. (2026, November 20). “GTA 6” raked in $3.8 billion sales on first day: The largest entertainment launch ever. Yahoo Finance. https://finance.yahoo.com/media-advertising/articles/gta-6-raked-3-8-200746469.html
Gee, J. P. (2003). What video games have to teach us about learning and literacy. Computers in Entertainment, 1(1), 20. https://doi.org/10.1145/950566.950595
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. https://doi.org/10.3102/003465430298487
IGN. (2026, November 18). GTA 6 review: One of the most impressive games ever made. IGN. https://www.ign.com/articles/grand-theft-auto-vi-review
Jordan, K. (2015). Massive open online course completion rates revisited: Assessment, length and attrition. International Review of Research in Open and Distributed Learning, 16(3), 341–358. https://doi.org/10.19173/irrodl.v16i3.2112
Kapur, M. (2008). Productive failure. Cognition and Instruction, 26(3), 379–424. https://doi.org/10.1080/07370000802212669
Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75–86. https://doi.org/10.1207/s15326985ep4102_1
Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press. https://doi.org/10.1017/CBO9780511815355
Marsick, V. J., & Watkins, K. E. (2001). Informal and incidental learning. New Directions for Adult and Continuing Education, 2001(89), 25–34. https://doi.org/10.1002/ace.5
Mashable. (2026, November 18). How long it takes to play “GTA 6”: Story playthrough and achievements. Mashable. https://mashable.com/entertainment/how-long-will-it-take-to-play-gta-6
New London Group. (1996). A pedagogy of multiliteracies: Designing social futures. Harvard Educational Review, 66(1), 60–93. https://doi.org/10.17763/haer.66.1.17370n67v22j160u
Poquet, O. (2024). A shared lens around sensemaking in learning analytics: What activity theory, definition of a situation and affordances can offer. British Journal of Educational Technology, 55(4), 1811–1831. https://doi.org/10.1111/bjet.13435
Reda Sadki (2012). Badges for online learning: gimmick or game-changer?. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/yhjxg-09z18
Sacnilk. (2026, November 19). GTA 6 breaks global livestreaming records with nearly 8 million viewers as launch sends internet traffic to new high. Sacnilk. https://www.sacnilk.com/news/gta-6-breaks-global-livestreaming-records-with-nearly-8-million-viewers-as-launch-sends-internet-traffic-to-new-high
Sadki, R. (2018). Why gamification is a disaster for humanitarian learning. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/sc0yp-mdk52
Sadki, R. (2024a). Calculating the relative effectiveness of expert coaching, peer learning, and cascade training. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/5ry62-hv582
Sadki, R. (2024b). How does the scalability of peer learning compare to expert-led coaching “fellowships”? Reda Sadki: Learning to make a difference. https://doi.org/10.59350/snw2h-c5f75
Sadki, R. (2024c). Making sense of sensemaking. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/esjf6-p4w76
Sadki, R. (2024d). Taking the pulse: Why and how we change everything in response to learner signals. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/dxbpy-yst93
Sadki, R. (2024e). Why does cascade training fail? Reda Sadki: Learning to make a difference. https://doi.org/10.59350/j8vg0-yng46
Sadki, R. (2025a). Against chocolate-covered broccoli: Text-based alternatives to expensive multimedia content. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/n1d17-7r990
Sadki, R. (2025b). Online learning completion rates in context: Rethinking success in digital learning networks. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/qadwd-87309
Sadki, R. (2025c). Richard Mayer’s research on multimedia for learning actually proves text works better. Reda Sadki: Learning to make a difference. https://doi.org/10.59350/j989x-zqq65
Siemens, G. (2005). Connectivism: A learning theory for the digital age. International Journal of Instructional Technology and Distance Learning, 2(1), 3–10. https://www.itdl.org/Journal/Jan_05/article01.htm
Siemens, G. (2013). Learning analytics: The emergence of a discipline. American Behavioral Scientist, 57(10), 1380–1400. https://doi.org/10.1177/0002764213498851
Steinkuehler, C., & Duncan, S. (2008). Scientific habits of mind in virtual worlds. Journal of Science Education and Technology, 17(6), 530–543. https://doi.org/10.1007/s10956-008-9120-8
Take-Two Interactive Software. (2026, November 20). Grand Theft Auto VI shatters entertainment box office records with $3.79 billion in day one retail sales [Press release]. https://www.take2games.com/ir/news/grand-theft-auto-vi-shatters-entertainment-box-office-records
TechRadar. (2026, November 18). GTA 6 review: The most complete open-world experience you will ever play. TechRadar. https://www.techradar.com/gaming/gta-6-review
Vice. (2026, November 18). GTA 6 PS5 trophy list revealed: How many trophies do you need for platinum? Vice. https://www.vice.com/en/article/gta-6-ps5-trophy-list/
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (M. Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds.). Harvard University Press. https://doi.org/10.2307/j.ctvjf9vz4
