The Dangers of AI, Microtransactions & Lootboxes | Design Dive

The Dangers of AI, Microtransactions & Lootboxes | Design Dive

Hi I’m Tommy Thompson and welcome to Design
Dive here on AI and Games. In recent years
a hot topic within the games industry is the
adoption of microtransactions, where players
spend are encouraged to spend money on a variety
of small add-ons that can change cosmetic
appearance of characters or outright influence
the overall experience. In addition, you’ve
had lootboxes, where you don’t know what item
you’re getting until you take the gamble and
open the box. Throughout this current generation
of gaming, we’ve seen a significant increase
in these monetisation models being employed
in large-scale commercial titles without appropriate
consideration, to a point that it received
significant backlash not just from players,
but it has now invoked the ire of governments
around the world and regulation is hot on
everyone’s mind. In this episode I want to
dive into this issue, why it’s so problematic,
but more importantly, the clear and present
danger that AI presents to consumers as it
is adopted in these monetisation systems.
Now before we get into the meat of this video,
I want to make two points. First let me stress
that what the points I’m making are not based
on evidence I have found of specific AI applications
to microtransactions and lootboxes in commercial
games, but rather an argument based on my
knowledge and understanding of this industry,
how game publishers aggressively pursue in-game
monetisation and how these machine learning
technologies can be employed in this space.
So while this is purely conjecture on my part,
it is an educated argument – or at least I
like to think it is – and I strongly imagine
that some of what I’m suggesting is either
on the horizon, if not already underway.
Secondly, as you’ll notice throughout this
video, my focus is on the behaviour of publishers,
not developers. Game developers often receive
the bulk of criticism and harrassment for
something they have little to no say in as
a game is being built. And that is frankly
unacceptable. In my experience developers
are – like players – often having to work
around the realities of having these microtransactions
injected into the games they’re involved with.
As such, if you have a bone to pick with the
topics of this video, you take it to the publisher
and more importantly, you speak with your
wallet. You don’t like what’s happening here,
don’t spend the fucking money.
So let’s go back to the beginning: microtransactions
are – as the name implies – an effort to have
players of a game spend small amounts of money
as part of the ongoing interaction and enjoyment
of the product. These small purchases often
are tied to elements that are cosmetic in
nature, such as costumes for characters, or
they can have some actual value in the context
of the game itself. This could mean extra
lives in Candy Crush or new characters in
a game such as Marvel: Contest of Champions.
The concept arose predominantly from mobile
gaming in late 2000’s, where instead of selling
games at a specific price point, they are
completely free to download, but your ability
to progress in these games and the rate at
which you do so is often heavily tied to the
amount of real-world money players inject
into it. This is certainly a popular and lucrative
format, but it requires players to continue
to invest their time and more importantly
their money, into the game itself, with this
leading to a variety of complex and convoluted
economies emerging in these games to not only
reinforce the need to spend money, but obfuscate
the true value of what you’re spending money
This has been compounded by the lootbox: an
idea that existed in mobile gaming for over
10 years before being popularised in the console
and PC gaming market by Overwatch published
by Activision Blizzard. In each instance,
players open a box that has one or more items
of potential value to players. These boxes
can typically be earned through in-game activity,
but at a relatively slow pace. However, they
can more easily be attained by spending money
within the games economy. In each instance,
the probability of a valuable item is often
incredibly low and only now are games being
forced in some markets to display the probabilities
of attaining the most lucrative items on offer.
All of these practices have continued to evolve
and mutate as they thrive not just in mobile,
but also in the console and PC-gaming spaces.
However, there has been some pushback as it
moves further into the console market, given
that these economies are being injected into
paid products. Players have already spent
say £40 or $60 on the game, with publishers
subsequently trying to squeeze more cash out
of their players. Much of the AAA game industry
– most notably Activision, Take Two Interactive
and Electronic Arts – has embraced this. While
the argument is often made it’s a reality
brought on by the ever increasing costs of
game development, many of these publishers
have seen revenues and in-turn profits skyrocket
largely due to in-game monetisation. One need
only look at Electronic Art’s FIFA and Madden
franchises: where the Ultimate Team component
of these games – where players purchase random
digital cards based on real world sports stars
– garnered $1.1 billion dollars of revenue
in 2018. That’s more than the boxed games
themselves and account not just for 50% of
all revenue from digital products by the publisher,
but is also 21% of EA’s total revenue for
that fiscal year. Publishers want microtransactions
to stay, because they’re making them a LOT
of money.
It’s been a huge point of consternation for
players given the impact it has on how games
are designed and the psychological manipulation
that becomes laced throughout a given title
in an effort to encourage you to spend money.
In a previous Design Dive on Dark Souls I
discussed how games build reward loops and
feedback systems to encourage continued play
and provide a sense of progression. While
this typically laced into player performance
and perseverance in the case of Dark Souls,
many economy-driven games either stifle or
limit that progression without a further injection
of cash or outright inhibit it. While many
publishers – and the players who play their
games – often defend these practices given
the additions found in microtransactions and
lootboxes are cosmetic in nature such as new
skins for characters or weapons, it is deliberate
choice that inhibits a players engagement
or enjoyment. Cosmetic items are often heavily
tied into the social element of play in these
games. You’ll note that most lootbox-heavy
games are ones with online multiplayer, whereby
you’re not just defeating the enemy team,
but you’re doing it whilst showing off your
cosmetic bling. This adds social stigma to
an existing pressure to engage in the monetisation,
given it creates fashion trends within the
ecosystem of the game. This might seem trivial
to some, but in the case of games such as
Fortnite, it leads to bullying in schools
towards kids who don’t spend money given they’re
‘default’ for wearing the free original skins.
Everywhere you go, monetisation-laden games
are deliberately constructed to steer players
towards in-game purchases, with lucrative
offers made early in a game to get you on
board. As well as periodic deals that encourage
you to maybe put down a few bob here and there
to get that item you wanted. This combined
with obfuscation of the in-game economy can
have a crippling effect on players and their
wallets. Particularly if the player in question
suffers from gambling addictions or if children
are involved. In-game monetisation is not
restricted in accordance with player age,
as such when games like FIFA or Crash Team
Racing are rating by the PEGI and ESRB rating
bodies as safe for children, they don’t factor
the monetisation systems involved. Plus in
some instances publishers such as Activision
frequently add the monetisation system as
an update to the game *after* it is rated
and released in an effort to avoid negative
press in the build-up to release. Thus further
adding to the confusion and surreptitiously
driving you towards spending money.
In addition to all this is the Battle Pass
approach, adopted in the likes of free-to-play
games like Dota2 as well as battle royale
games like Fortnite and Apex Legends, but
they are rather upfront about what they’re
doing, given it’s clear what you’re paying
for as well as what you can attain. However,
they’re still a bit nasty, given they are
essentially a purchase that requires a time
commitment, given that the unlocks within
a battle pass are often not available after
the original gameplay season that pass is
associated with.
Now in late 2019 we’re seeing this begin to
reach a critical juncture. As publishers brazenly
slap more and more microtransactions on premium
$60 products aimed at children without fear
of reprisal, gambling commissions and governments
around the world are now looking to intervene.
While some countries have outright banned
lootbox behaviour given the principle of spending
money (albeit indirectly) for a random reward
is tantamount to gambling, others such as
the United States and the United Kingdom are
now investigating what new forms of legislation
could and should be implemented to protect
vulnerable individuals from what is an otherwise
predatory behaviour. Resulting in the likes
of EA and Epic Games being brought forward
to speak at committee level of the UK government.
This is a pretty scary proposition as EA act
like a bunch of chancers by drawing parallels
to the likes of Kinder Eggs and collectable
card games in an effort to normalise their
behaviour as ‘surprise mechanics’, which is
– professionally speaking – a load of old
horse shite.
So why am I talking about all of this? Two
words. Player analytics. It’s a process where
you gather up data en masse about the behaviour
of players within your game in order to better
understand what they’re doing and how they’re
going about it. Over on the main AI and Games
series, I’ve explored analytics a couple of
times already, with the likes of the PsyOps
project being able to establish how players
adopt specific gameplay responsibilities in
Battlefield depending on their personality
or more critically, their age. Meanwhile the
video exploring research in Tomb Raider: Underworld
highlighted one of the first real efforts
by developers to understand what is happening
in their game after it’s released into the
wild by reading data from players over the
Xbox Live network. But perhaps more critically,
is the episode from back in 2017 where I explored
academic research that examined the Team Fortress
2 hat economy. As explored in the episode,
there are distinct correlations that can be
estavlished from perceived social status in
the games ecosystem to the quality of virtual
items they possess and the distinct profiles
of players that exist – ranging from those
simply uninterested to those who crave attaining
the most lucrative items. Perhaps most critically,
is the research itself warns of sociological
issues that can emerge when social dynamics
intersect with virtual economies, with the
very real threat of privilege and marginalisation
emerging in these communities.
Now all of these research projects are comparatively
quite old now. The Team Fortress and Battlefield
research date back to around 2012-2013, while
the Tomb Raider Underworld project dates back
all the way to 2009. Back when originally
published, the idea of recording player data
to gather fairly accurate information about
behavioural traits was very much in its infancy.
Meanwhile in 2019, any publisher working on
a live-service games will be have entire data
science divisions dedicated to player analytics
and other heavily data-driven fields. It’s
an entire facet of game development that has
only really emerged in the last 5-10 years.
These teams can be focussed perhaps on individual
titles or all titles under the publisher.
Analytics of player data is equally applicable
to your behaviour around monetisation as it
is for any other in-game activity. Now this
has been happening – especially in mobile
games – for a few years now. Such as identifying
the types of products people buy and whether
particular levels of a game are where most
players get stuck, given it provides an ample
point to sell a booster pack at a discounted
rate. But there are already existing machine
learning driven metrics regularly used when
assessing a players engagement in a particular
game. One of the most well known examples
is called churn prediction.
Churn prediction is used in a variety of consumer
services on the web and it’s used to identify
when you – the customer – become a churner,
meaning you’re not going to continue to engage
in a particular service and – more critically
– won’t pay them for it. Hence Amazon goes
out their way to send you emails if you haven’t
bought anything in a while, or Netflix might
encourage you to try out a new show if you
haven’t used your subscription and they’re
worried you’re not going to renew. All of
this in an effort to put your money back into
their ecosystem. Now churn prediction in games
is identifying two key things related to players
and microtransactions:
– First of all identifying the ratio of churn
within the game. Where by examining in-game
behaviour of all players, you can understand
what proportion of the audience will spend
money on microtransactions within a period
of time, if ever.
– Secondly, the churn rate: at what point
do players turn into churners.
Churn predication can be achieved using a
variety of machine learning technologies.
Once you combine this with player segmentation
– where you cluster and categorise players
based on specific attributes, ranging from
in-game performance to your engagement with
the monetisation services, then you’re getting
to the interesting but also scary point. At
an abstract level the entire playerbase can
be analysed to see what’s happening: where
are the problems in the game, what balancing
issues need to be addressed, but also what
monetisation packages are proving popular
and how many players are not spending money.
Now I don’t like to be doom n gloom about
AI on this show and I rarely ever am. In fact
AI and Games was built to challenge the stigma
and bad reporting that surrounds artificial
intelligence. But the problem is, AAA publishers
are really embracing this stuff and they don’t
have a great track record with monetisation
as it is.
Right now we’re in a position where at a top
level decisions can be made to rebalance and
tweak parts of a game and in the context of
monetisation, edits can be made to the economy
of a game such that things become easier or
harder. Meanwhile at present, new promotional
products can be released into the economy
and targeted towards players who have never
spent money or spend very little money. This
is a tactic that already exists within mobile
gaming, given it’s a little easier to make
these changes so quickly and that it’s appealing
to a broader base of consumers that are a
little less incensed by it all.
Now, take this to the next logical step – this
is the big conjecture part btw – where your
machine learning prediction systems become
ongoing real-time services. Where a game can
poll these analytics services, to make judgements
on what changes to make to the game you’re
playing, to encourage you to spend money,
in real-time. Said game is tweaking its economy
and progression systems and even coming up
with unique and bespoke products – for you,
this particular player based on your current
gameplay experience. And it’s being driven
by the analytics services that have been built.
Does this sound scary? Yes! Is it far fetched…
I don’t think so, otherwise I wouldn’t be
talking about it. Don’t forget in 2017 Activision
was granted a patent based on ideas of manipulating
matchmaking to encourage use of microtransactions.
Be that to give you an advantage in the game,
or just to force you to play against opponents
with cooler outfits or costumes. How do you
think such a system would be able to identify
what brackets you fall into for the matchmaking
process? This, right here, is what I’m worried
Imagine games are being tailored to appeal
to both casual and more hardcore players,
with the in-game monetisation prioritising
specific aspects of player behaviour within
those sectors. Perhaps for the more hardcore
is offering better deals on cosmetics, while
the more casual audiences are getting better
affordances for progression given they want
to avoid a grind that’s being artificially
inflated in real time? I can see this happening
in mobile given as a I said, mobile audiences
aren’t typically not as actively engaged in
the media that surrounds a game. Plus many
of these games are actually isolated individual
titles without a larger multiplayer meta that
players a communally trying to establish.
And there are products out there offering
machine learning driven analytics for mobile
developers. Again, I recognise the value of
it, but all it takes is for one stupidly popular
game to figure out how to pull it off.
Naturally I’m worried about my hobby (or whatever
my relationship to games has become in recent
years). But this could be a real red-flag
that could impact more vulnerable individuals
who are skirting around games that have been
soaked in microtransaction models. There needs
to be a better understanding of what these
systems can do and how they’re being employed
in particular games. Would legislation help?
Perhaps. But governments are often slow to
understand how technology works and then make
the most heavy-handed and ill-judged decisions.
This is why we’re in this situation to begin
with, given nobody really noticed the effect
all of this is having on games and their players
until now. Plus if you consider my two biggest
audiences on this channel – the US and the
UK – I don’t think either country can really
advocate for how clued-in their governments
are when it comes to protecting the wellbeing
of their citizens from technological upheaval.
Is this all being blown out of proportion?
I’d like to think so, but on the other hand
this is an idea that I can imagine investors
are salivating at the prospect of. Right now
there is next to no consumer protection or
appropriate guidance given when using these
sorts of practices. Particularly lootboxes.
Only now at the time of writing this video
did news break that platform holders will
start to release the odds of lootboxes, several
years after they’ve flooded the market. It’s
a start, but I don’t think it’s going to be
Either way, I think it’s real food for thought
and while this is a bit of a detour from regular
episodes of Design Dive, I think it’s something
that I needed to talk about! But don’t sweat,
I’ll be back talking about video games in
the more traditional sense in the next Design
Dive episode. Be sure to subscribe as after
two years I’m returning to Dark Souls next
month as I talk about my time playing Dark
Souls II – Scholar of the First Sin as I continue
my rather personal journey through the series.
That’s it for this episode of Design Dive
– part of my work here on AI and Games that
is sponsored by these wonderful folks you
see on screen now. Join the AI and Games Patreon
or become a member of the YouTube channel
to help support my work. Thanks for watching,
stay safe, have fun, I’ll be back.

16 thoughts on “The Dangers of AI, Microtransactions & Lootboxes | Design Dive”

  1. I'm not one for making doom'n'gloom predictions on this show, particularly when it comes to AI. But my issue isn't with AI, it's with corporations and their rather greedy behaviour. Machine learning is already being used in some corners of in-game monetisation, but there are some very real and very scary possibilities that could arise if publishers think they can pull it off. So this Design Dive episode is actually tying into a lot of existing videos on the show as I look at the logical conclusion of where this research could be applied in AAA games.

  2. It's so weird to me when creators bleep their own swear words in videos. I can't quite put my finger on why, but it annoys the crap out of me! I'm not prudish — by all means swear! — but to choose to swear KNOWING that you're going to censor yourself (presumably because you don't want to be demonetised) seems at odds with the purpose. Employ some wit when you're writing the script and come up with some cutting turn of phrase to emphasise your point instead and it'll end up making your point stronger than some neutered 'beeeep'.

  3. If you can't control yourself from throwing money into a bottomless pit , not only do you not deserve my sympathy but also society should not do anything to protect you. Let Darwin do it's damn job.

  4. Sorry man, but the developers deserve every bit as much shit as the Publishers, if not more. You're obviously not going to agree with me, but, the Publisher may mandate it, the developer has to gamify, develop, and implement that. Bioware can eat shit all day, in and out, and they should hear about it. So should DICE. So should Blizzard. Valve doesn't get off either, but at least they took a stand against roulette websites, but that doesn't mean they aren't Innocent and undeserving of as vocal criticism as the rest. Their little card game was a very profound lesson for them they took squarely on the chin. But before that, people made it very clear they weren't for that garbage. That doesn't mean harassment, that doesn't mean calling them names, or threatening violence on them, but the critical voices should be heard echoing even throughout the bathroom stalls by everyone. Its disgusting behavior, and just like those egging them on to do it, the people creating this garbage are culpable for the problems they belong too.

  5. Hey, thanks for discussing this in what's normally a more mechanical based show. It's super important and it's nice seeing a person who has the clout of being a game dev really talking about this as a problem. Mostly whenever you hear microtransactions talked about in game media and journalism, it's discussed as a necessary evil and not as harmful. Or at least, that's what the discussion was until the hammer started coming down on them.
    Seriously, thank you for this and for all the work you do making videos.

  6. 2 minutes in, but I take some minor issue, namely: Voting with your wallet I object a little too. I agree it's important, but is not the be-all-end-all as people will be tempted one way or another. It's like telling people to stop the drug trade by voting with your money. I can't stop my cousin from spending their livelihood.

    As for blaming the developers, I hope for a little more clarification. Do developers get the say in how to balance their game, stretch out the grinds, etc., or the publishers? If the developers are free to balance it how they want, they can be at fault in-those-cases, unless you want them voting with their wallets by letting it starve a little by protesting against putting lootboxes/microtransactions in. Though that is a whole other moral discussion of 'Do you blame the soldiers of war crimes when they're ordered to do it?'

  7. I agree with everything you said and your analysis of the issues at play is spot on… The unfortunate reality is that by the time enough people realize what's happening the damage has already been done.

  8. Man I haven't spent one cent on this bullshit.

    But somehow my, less than stellar grown up friends, will spend $15 ON A FUCKING FORTNITE SKIN. Motherfucker, you can buy both metro 2033 and last light games and a fucking hoagie for $15.

    These ADULT friends of mine have spent hundreds on just skins. I can understand kids and peer pressure. But grown men and women.

    I stopped playing games like Fortnite after one season since my friends stopped begging me to play.

    As you can tell, my grammar shows my frustration.

  9. Great video, on Overwatch and Lootboxes, yes the fastest route to getting lootboxes is to purchase them, however Overwatch in my experience has had the most generous and non-predatory lootbox system to date. You get up to 3 per week by winning 9 arcade mode matches (3 wins = 1 Lootbox), you get 1 every time you level up, which requires 20k XP earned; I generally get anywhere from 2000-5000 XP per game, it isn't a huge challenge to level up, and on top of this about every two weeks or so you are given a number of lootboxes based on your endorsement level. Endorsements (such as Good Teammate, Shotcaller, Sportsmanship, etc.) can be given to any of the other players after every match, the more you get, the higher your endorsement level, up to 5, you have to continue receiving endorsements to maintain your level. Level 1 gets 0 because they've not been endorsed at all or very little to not reach level 2. Level 2 gets 1 lootbox, and so on where level 5 gets 4 lootboxes every few weeks. I've personally never even seen level 5, and only know of the existence of it through Reddit, but I've maintained level 4 since the system was introduced.

    TL;DR: Overwatch has the most generous lootbox system by far, guaranteeing at least 4 lootboxes per week (you will undoubtedly level at least once throughout the 9 arcade wins required to earn all 3 lootboxes for the week) with the potential to earn another 4 based upon being a good, friendly teammate, AND a very consistent rate of earning them individually by just playing the game through leveling.

  10. Even as a prediction everything. Literally EVERYTHING here is not far off. In fact it's already happened and has been reported by many others experts gamers and journalists. Even ESA is somehow in on this with their data breach leaking personal information of any Youtube influencer or gaming journalist who has been to E3. Just over the weekend one journalist was under a private investigation from Take two studios for a leak on one of their projects.
    What this means is that now the game publishers with the help of this dysfunction can police anyone as a game journalist from revealing anything about their game that they consider to be sensitive information. Now this policing can even come in the form of threats to not reveal any information about a game. This may include anything from whether or not the game has multiplayer or season passes … Or if it even has lootboxes.

  11. A gaming crash will cleanse this problem. All we gotta do is make sure the government gets a knife (regulation) in the back to loot boxes before it happens so when games do get out of the crash like last time, we'll have a lootbox free hobby again.

  12. I attended a lecture by a Zynga data analytics guy about 4/5 years ago. All the points you make about letting AI detect if someone is likely to leave the game and offer them a great deal for pennies is one of the things he was telling folks. "It doesn't cost you anything, and if they respond to it you made a little money. They were going to leave anyway."

    Jim Sterling had a Jimquisition with some backend "players into payers" service that would do the same, but it would automatically adjust prices for every user to try to maximise how much money you could get from them.

    I would honestly be more surprised if the big publishers didn't try to put this stuff into games. It was a great talk, really opened my friend's eyes to just why 'big data' was such a big buzz, and confirmed the shady side of big developers/publishers for me.

Leave a Reply

Your email address will not be published. Required fields are marked *