EAI Endorsed Transactions

on Context-aware Systems and Applications Research Article

How instability in virtual economies of mobile digital

games drives and ruins profit
P.C. Lohse1,*

1
Andrássy University, 1088 Budapest, Hungary

Abstract

The digital gaming business has changed in the last years. Digital games are no longer just products. They developed into
services. People play games not only on a stationary device, but on mobile devices, too. The change of gaming devices
also had an impact on game design. Competition between mobile digital games mostly takes place at a download price of
zero. Today the Freemium monetization method is the dominant monetization method. Players of digital games often have
to go through a core loop by repeating the same tasks inside the game. The stability of the virtual economy inside the game
affects the game’s core loop. This paper discusses the role of instability in virtual economies in the context of core loops.
A small instability inside a virtual economy can increase revenue of a digital game. Players tend to stay in a game because
of sunk cost fallacy. If instability inside a virtual economy happens for a too long time players will quit playing and
revenue will decrease.

Keywords: Freemium, mobile gaming, sunk cost fallacy, instability in virtual economies, core loops

Received on 04 November 2019, accepted on 20 December 2019, published on 10 January 2020

Copyright © 2020 P. C. Lohse, licensed to EAI. This is an open access article distributed under the terms of the Creative Commons
Attribution licence (http://creativecommons.org/licenses/by/3.0/), which permits unlimited use, distribution and reproduction in any
medium so long as the original work is properly cited.

doi: 10.4108/eai.13-7-2018.162632

*
Corresponding author. Email: [email protected]
revenue had a price per download [7]. Players can
download the applications for free and test the game. If
1. Introduction they like it, they can get an advantage in the game by
purchasing a virtual currency or other premium features.
The way people play mobile digital games has changed. The game publishing companies do no longer need a high
Mobile digital games became more popular over the last amount of downloads to achieve high revenues.
years [10]. Mobile digital games are no longer a product. Competition between game publishing companies for new
They developed into services [10] [12]. Mobile digital users mostly takes place at a download price of zero [7].
games with the highest revenue in the apple app store, Today revenue of digital mobile games is driven by in-
Google Play Store or others often are available for free app-purchases. An in-app-purchase is a purchase that is
[7]. The majority of mobile digital games are available done inside the application to use special premium
without a price per download and make money via in-app features. In July 2016 99 % of the top 200 applications by
purchases [7]. Competition between digital mobile games revenue in the Apple App-Store offered in-app- purchases
mostly takes place at a download price of zero. Players [7].
pay for premium features inside the games. This Users of mobile gaming applications often repeat tasks
monetization method is called Freemium [5]. Today inside the game. The game starts a trigger, which leads to
Freemium is the dominant monetization method for an action of the player. The game rewards the player for
mobile digital games [6]. completing the action. This reward can be invested by the
In July 2016 98.5 % of the top 200 Apple App-Store player so that he can have an advantage in the next run.
applications by revenue were available for free [7]. Only Eyal called these four stages a core loop[9].
1.5 % of the top 200 Apple App-Store applications by

EAI Endorsed Transactions on

1 Context-aware Systems and Applications
Online First
P. C. Lohse

Mobile gaming applications sometimes take advantage of is discussed in the book “Online Worlds: Convergence of
dark gaming patterns and behavioral economic theories to the Real and the Virtual“ by Bainbridge et al.
motivate players to buy premium features inside the game In 2007 Friedman et al. published a scientific paper
via in-app-purchases [2] [3] [4] [8] [9]. The focus of this about sunk cost fallacy. Their paper is about a laboratory
paper is the role of instability in virtual economies. The experiment. In their experiment they used a computer
over expansion of a virtual hero’s traits leads to instability game that confronted players with different variations of
of a game’s virtual economy. sunk cost.
Players run through a loop while playing a mobile gaming In 2014 Dimitar Draganov has published his book
application[9]. Game publishing companies can set up a “Freemium mobile games - Design & Monetization“. In
virtual economy inside a game in a way so that people his book he discussed how game design can keep people
have to run through this loop more often or spend more interacting with a digital game over a longer period of
money. If players have to run through this loop too fast time and how to convert players of a game into paying
and too often to compete with others they might stop customers.
playing the game. Sunk cost fallacy might be a reason B. F. Skinner has laid the behavioral economic
why they continue playing even though they don't like the foundation for variable payoffs with his research into
game anymore [3]. This contradiction will be discussed in operand conditioning. His research indicates that a
a case study of the mobile gaming application "War of variable payoff is a stronger motivator for people than a
Nations". "War of Nations" is a mobile gaming predictable payoff. B. F. Skinner was known for his
application that was successful in many different experimental setup in a box, which he used to research
countries some years ago. Over the years the game has operand conditioning. This experimental setup has been
lost it's top position within the apple app-store. This paper included in literature under the term Skinnerbox. Variable
will have a closer look at why the decline of the ranking earnings are common in digital games. This can be a
for this particular mobile gaming application happened. motivation for playing a game longer.
After a literatur review, the research question will be N. Eyal has discussed the concept of the core loop in
introduced. To answer the research question, a theoretical his book "Hooked". Core loops are composed of the four
framework and the role of commanders inside the game phases trigger, action, (variable) reward and investment.
need to be discussed. The theoretical background will lead A player goes through these phases when he is playing a
to research method and possible biases. A first view of the game. According to Eyal, these four phases are behavioral
collected data is presented in the chapter descriptive and can be used to allow a player to interact with a game
statistics. Based on the theoretical framework and the for as long as possible.
collected data a hypothesis will be formulated and a
regression analysis will verify it. In the last chapter
conclusions and further research approaches will be 3. Research question
discussed.
Most of the relevant literature discusses the behavior of
humans in digital games and virtual economies on a
2. Literature review general level. Only a few studies like Friedman et al in
2007 published about specific experiments and not on a
Zagal, Björk and Lewis published a paper on "Dark wide unspecific general level. This paper focuses on an
Patterns in Game Design" [8]. In their paper, the authors empirical case study. The advantage of an empirical case
deal with dark game patterns. Dark game patterns are study is that conclusions can be made not only on a
defined by the authors as a negative experience for the specific level. The presented research in this paper is
player, which is intentionally generated by the respective close to a field study.
game company. Zagal, Björk and Lewis define monetary The central research question is, how instability in the
dark game patterns in a way that any in-game purchases a virtual economy of the digital mobile game "War of
player regrets are caused by dark gaming patterns. In dark Nations" has affected the revenue over time.
gaming patterns the player loses track of how much
money he has already spent. If the player is not sure about
what exactly he is buying or how much he has to buy to 4. Theoretical framework
reach his respective target within the game, then there
may also be a dark game pattern. Games are no longer Players often develop an emotional connection to a
just products. They changed to services. People use games game. This emotional connection is not only determined
over a longer period of time and the game publishers keep by the fun a player has from playing the game. The longer
updating the game. In 2014 Oscar Clark has discussed this a player plays a game and the more money a player
phenomenon in his book “Games as a service. How free spends on a digital mobile game the more rational it
to play can make better games“. becomes to him to spend even more time and money on
There is research about similarity between real and this game. The more time and money are spent on a game
virtual worlds. The field of human- computer interaction the higher are the player's sunk costs. Quitting a game will
confront the player with the maximum amount of sunk

EAI Endorsed Transactions on

2 Context-aware Systems and Applications
Online First
 How instability in virtual economies of mobile digital games drives and ruins profit

costs. It is rational for a player who already is at a higher

level in a digital mobile game to continue playing this
game even if another game would give him a higher
amount of happiness. These players consider sunk cost
and have an emotional connection to the game because
they build up their own account and already have
achieved success in a game. The longer players play a
digital mobile game the harder it becomes for them to
abandon this game. Some players will continue to play
even if they don't like the game any more. Their choice is
between two negative payoffs. These choices are the sunk
cost of abandoning the game and the cost of staying inside
the game.
In behavioral economics this effect is called the sunk
cost fallacy. Sunk cost fallacy is an effect that describes
the choice between two options with different sunk cost.
Only one option is possible to be taken. It often seems to Figure 1. Sunk cost bias (own figure based on
be rational to people to use the option with higher sunk Kahneman, Knetsch and Thaler)
cost. Even if the option with smaller sunk cost gives the
person a higher value. Staying inside the game will
increase the sunk cost for the player. The incentive to stay Transferring the sunk cost bias in the context of sunk
inside the game increases the longer a player already cost fallacy in mobile digital games, this leads to the
plays the game or the more money he already had spent. result that a small amount of instability inside a virtual
If a player is confronted with loss in value of his economy is needed to increase profit. A small instability
account there is a huge stimulus that a player, who already inside the virtual economy of a digital mobile game
spent a longer time playing a game or already spent confronts players with a disproportionately amount of
money on a game, will increase his ambitions inside a loss. Players loss aversion can increase profits of game
game. Loss in valuation of a gaming account can occure publishing companies. The higher the instability inside of
because of events like instability inside a virtual economy. a virtual economy the higher will be the loss the players
Instability in a virtual economy can happen in many are confronted with. By increasing the time spent on the
different ways. One example is the announcement of a game the players are confronted with higher sunk costs.
new very strong hero. If the new hero is a lot stronger The game design only has to trigger the loss aversion of
compared to previous heroes, he makes these older heroes the players. Loss aversion leads to a phenomenon called
completely worthless. Players will have to get this new “sunk cost bias“. A loss of the players achievements can
hero as fast as possible to have an advantage over other happen due to instability in virtual economies. The
players. In the medium run all players will need this new success a player has achieved is reduced compared to the
hero to stay competitive inside the game. moment the instability started. The prospect of loss is a
The stimulus of loss happens because of sunk cost bias. more powerful motivator for peoples behavior than the
Usually an additional gained unit gives the user a positive prospect of a gain. The overvaluation of loss can stop
increase in utility valuation. The marginal utility unit is people from quitting the game. This can lead to sunk cost
positive, but decreases as shown in the following figure. fallacy. People have the choice between quitting the game
Smaller gains are relatively higher valuated than larger or spending even more time and / or money on a game.
gains. A loss in utility should be the same, but it isn't. This can conduct to higher profits for companies. It
There is an asymmetry around the neutral point. People in becomes profitable to confront players with loss.
general overvalue losses compared to gains. This is shown Instability in a virtual economy is a way to do so. Players
by the red line in the figure 1. can continue to play or quit the game. Quitting the game
means the maximum amount of sunk costs to the player.
Players will avoid this option as long as possible. Some
players will even continue to play, while not having fun,
but to avoid negative emotions. This is a social
undesirable status.

5. The role of commanders inside the

game
War of Nations was first released in June 2013. In this
mobile digital game players compete against each other or
in teams. Players build up an army and bases. There are

EAI Endorsed Transactions on

3 Context-aware Systems and Applications
Online First
P. C. Lohse

different units. Each kind of unit has a certain unit value. iPhone and ipad. For each quarter an average rank has
Buildings are inside bases and can be levelled up to a been calculated. There is no public data available on how
maximum level. Players hit other players’ bases with their large each countries market for “War of Nations“ actually
army. They try to win temporary events inside the game is. Therefore no weighting for given countries has been
and earn rewards. These rewards are usually set up to done. The average sum of commander traits per quarter
increase the players’ or teams’ strength and give the has been calculated based on the data from social media
player or the team an advantage in future events. Players and forums[12] [13] [14] [15]. The rankings are indicators
and teams also compete in battlepoints. Battlepoints are for the popularity and the revenue of the game. The
the sum of total unit value destroyed by a player or his change in commander traits compared to the previous
team. quarter is an indicator for the loss aversion the game tries
Commanders are virtual heroes and the most important to confront the players with. The stronger the current
type of boost for the player's army. Each commander has commander gets, the weaker older commanders become.
traits for attack, damage and leadership. The stronger the There are unobserved variables in the following
commander's traits, the stronger is the player's army while regression analysis. There is no public data available on
attacking or defending. There has been a significant when a sale happened inside the game. This might have
increase in the strength of available commanders in the an impact on the game’s top grossing ranking. Sales
mobile digital game "War of Nations" over the last years. within other games might have a negative influence on the
Players need to collect experience points for each game’s top grossing ranking.
commander. This is how commanders can reach higher Other economic factors are not assumed in the model,
levels. Commanders of the same kind can be merged. The because the following model is using global average data
stronger the two merged commanders are, the stronger the for the game’s top grossing ranking. Macroeconomic
new commander becomes. The level of the commander variables should be added to the model in case that the
increases by merging commanders up to a maximum ranking for different countries is compared. This is not the
level. In the relevant literature this mechanism is called case in this study.
“complete gacha“. This means that players actually need a Based on these information the following research has
specific number of commanders of the same kind to get been done. Player boosts also affect commander traits and
one really strong commander that can reach the maximum since stronger players are more likely to post information
level. Players need to earn these commanders by winning of their commanders there might be an upwards bias in
events or making in-app-purchases. the data.
In this paper, the average sum of commander traits per
quarter is used as a proxy variable for instability inside
the game’s virtual economy. 7. Descriptive statistics
The collected data are presented in the following
6. Research method and possible biases figure. The left y-axis of the figure 2 shows the ranking in
the category “Gaming Top Grossing“, which is an
The company behind “War of Nations“ was asked on indicator for the revenue relative to all other competitors.
email and many other ways to support this academic The right y-axis of the figure 2 shows the commander
research by publishing statistics about commander traits. traits in 100k units. Data on commander traits are
They did not respond. To get an estimate about the available from 2nd quarter of 2014. On the x-axis the
increase of commander traits social media and game quarters from the 2nd quarter of 2013 until the 4th quarter
specific forums have been looked up for information of 2018 are shown. Each point in figure 2 gives the
about commander traits[12] [13] [14] [15]. The earliest
data about commander traits have been found from the
2nd quarter of the year 2014. This research focuses on the
time from 2nd quarter of 2014 until the 4th quarter of the Rica, Croatia, Cyprus, Czech Republic, Denmark,
year 2018. Dominican Republic, Ecuador, Egypt, Finland, France,
Based on the data about each commander's traits Germany, Ghana, Greece, Honduras, Hong Kong,
average sums per quarter from the 2nd quarter of 2013 Hungary, India, Indonesia, Ireland, Israel, Italy, Jamaica,
until 4th quarter 2018 have been calculated. The ranking Kazakhstan, Kuwait, Lao Peoples Democratic Republic,
is an average out of all available countries from Latvia, Lebanon, Luxembourg, Malaysia, Malta, Mexico,
appfigures2 in the top grossing gaming categories for Namibia, Netherlands, New Zealand, Nigeria, Norway,
Pakistan, Panama, Papua New Guinea, Peru, Philippines,
Poland, Portugal, Qatar, Romania, Russia, Saudi Arabia,
2
List of all available countries on appfigures with data for Singapore, Slovakia, Solomon Islands, South Africa,
“War of Nations“: Angola, Arabic Emirates, Armenia, Spain, Switzerland, Thailand, Trinidad and Tobago,
Australia, Austria, Azerbaijan, Belgium, Bermuda, Tunisia, Turkey, Tuvalu, Ukraine, United Kingdom,
Botswana, Brazil, Brunei Darussalam, Bulgaria, United States, Venezuela, Vietnam
Cambodia, Canada, Cayman Islands, Colombia, Costa

EAI Endorsed Transactions on

4 Context-aware Systems and Applications
Online First
 How instability in virtual economies of mobile digital games drives and ruins profit

average per quarter for the iphone / ipad category or the virtual economy took over and the ranking started to
average sum of commander traits per quarter. decrease.
The introduction and growth phase can be seen from The players had to go through the game’s core loop
the 2nd to 3rd quarter of 2013. From the 4th quarter of 2013 more often than they felt comfortable with. This means
until 2nd quarter of 2014 the maturity phase of the product that even though there is a sunk cost fallacy over time
life cycle happened. After this the decline happened. The players will quit the game, because the new virtual
decline stabilized, when the average sum of commander commanders are a lot stronger than previous virtual
traits started to increase. The average sum of commander commanders. The new game deciding content deflated the
traits is the proxy variable for the game’s virtual subjective value of older virtual commanders. This
economy. In the 2nd quarter of 2016 a big decline in the process should be able to be seen in a decrease of the
game’s ranking occured. The big decline happened nearly ranking in the category "Gaming Top Grossing".
a year after the instability started. After the huge drop the
ranking stabilized between rank 350 and 450.
9. Regression analysis

The question that has to be answered is, how the impact

of the average sum of commander traits per quarter to the
average ranking of the mobile gaming application "War of
Nations" per quarter in the top grossing category changed
over time. There will be a significant change in ranking,
which happened once in the data. To model this one time
change, the functional form is assumed to be like this:

Ranking is the dependent variable. Commander is an

independent variable for the average sum of commander
traits per quarter. Commanderhigh is a dummy variable
that equals 1, if the average sum of commander traits is
above 100.000. Commanderhigh is the variable that gives
further information about the one time effect of a high
Figure 2. War of Nations: Average Ranking of all average sum of commander traits on the ranking. The
available countries and average sum of commander value of 100.000 for high commander traits as a dummy
traits per quarter (own figure, data based on variable has been chosen, because this value gives the
appfigures[1], social media and game forums [12] highest R2.
[13] [14] [15])
If there is a negative effect of commander traits
towards the ranking of the mobile gaming application
8. Hypothesis "War of Nations" in the top grossing category the
independent variable for Commander and
The previously discussed leads to the hypothesis that Commanderhigh need to be positive. Both have to be
an increase in commander traits might have happened too statistically significant. The dummy variable
quick in the past of the mobile gaming application "War Commanderhigh shifts the intercept and gives information
of Nations". A short term instability might lead to an about when the effect of the average sum of commander
increase in revenue in short term, but a long term traits per quarter has an even stronger impact on the
instability of a virtual economy leads to a decrease of ranking. To avoid serial correlation generalized least
revenue. squares has been used. The regression result for the top
Players of mobile digital games tend to be happy about grossing iPhone category is the following:
the introduction of new contend to the game. This usually
leads to a positive impact of new content to revenue. The
positive effect of instability in the virtual economy of the
mobile gaming application “War of Nations” happened
between the 3rd quarter of 2015 and the 1st quarter of
2016. Although there was a positive impact of instability The R2 is 95.03 and the adjusted R2 is 94.75. The P-
in the virtual economy a better ranking couldn’t be values are 0 for Commander and Commanderhigh. Given
achieved, but the ranking stabilized. From the 2nd quarter a 5 percent significance level, this means that there is a
of 2016 the negative effect of instability in the game’s

EAI Endorsed Transactions on

5 Context-aware Systems and Applications
Online First
P. C. Lohse

statistical significant effect between these two 10. Conclusion and further research
independent variables and the dependent ranking variable. approaches
Some mobile digital games are designed to strengthen
effects of loss aversion and sunk cost fallacy. An
exploitation of these effects can happen by a high
instability inside a virtual economy. This instability can
last for a short, medium or a long time period.
A high instability in a virtual economy is a way to
exploit players. The ranking stabilized during the first
quarters after the introduction of the instability inside the
virtual economy. If instability in a virtual economy
happens for a longer period of time or increases too high
the game looses it's players. Players are the customers of
game publisher companies. Customers might get excited
about new content in the first moments, but adding more
and stronger game deciding contents will not increase
customer satisfaction in medium or long run. In the
medium or long run, there is no other way for players to
react to things like a high instability in virtual economies
than quitting the game. This means the maximum amount
of sunk costs for the players. A socially undesirable
situation for some players can be the consequence,
because some players will avoid to quit playing as long as
Figure 3. Regression plot quarter (own figure, data possible. For some players this can mean to continue
based on appfigures[1], social media and game playing without satisfaction.
forums [12] [13] [14] [15]) A small amount of instability inside a game’s virtual
economy might be necessary to keep a game interesting
for new players. Mobile digital games usually reward
The figure 3 shows the regression function using staying in the game for a long time. The longer players
generalized least squares. The moment the proxy variable continue to play the stronger their accounts get. This can
for the game’s instability in the virtual economy rises too make it hard for players to compete inside a game with
high the ranking drops. In this functional form the slope players who play a game for a longer period of time. A
before and after the drop in ranking is the same. small increase in commander traits and the overall virtual
If the average sum of commander traits increase by one economy can help to make old rewards useless after some
unit and all other factors stay constant, the ranking in the time. In this way a small increase compared to previous
category “Gaming Top Grossing“ increases by time periods inside a virtual economy can help to balance
0.0004188. This means that the number of the ranking a game.
gets higher. The game gets a lower ranking in the The presented model in this study has limits. The effect
respective stores. When the average sum of commander of the commander variable might be higher in one country
traits per quarter reaches 100.000, the intercept increases compared to other countries. Unobserved variables might
by 136.4343, because of the dummy variable have an impact on the ranking, too.
commanderhigh. Instability in virtual economies is a field that deserves
The hypothesis can be confirmed. Players of the further research. There is a research gap at the moment.
mobile digital game “War of Nations“ stayed inside the Game publishing companies can take advantage of their
game’s core loop for three quarters after the instability customers by confronting them with loss. One important
inside the virtual economy occured. Temporary positive question that needs more researchers' attention is how
effects can be interpreted. The downwards trend of the huge the socially undesirable effect for customers of
gaming application could be slowed down for three mobile digital games is.
months. In 2016 the negative effect of instability in the At the moment publishers of mobile digital games are
virtual economy started and the game’s ranking dropped. not faced with many regulations. There are discussions
The drop can be seen in the dummy variable about regulating lootboxes. In addition to the discussed
commanderhigh. topics in this papers there are a lot of asymmetric
After three quarters players anticipated the introduction information in the business of mobile digital games.
of new stronger commanders inside the game and the These asymmetric information are worth a further
devaluation of previous game deciding content. The discussion, too.
change in players anticipation about new content is the Corporate social responsibility can be a solution to
central reason why the drop in ranking occured. increase customer satisfaction and reduce socially
undesirable situations for customers. A code of conduct

EAI Endorsed Transactions on

6 Context-aware Systems and Applications
Online First
 How instability in virtual economies of mobile digital games drives and ruins profit

for human game design can be a step in the right direction

to avoid a socially undesirable status for players. A
limitation of instability in virtual economies of mobile
digital games can be a part of such a code of conduct. An
exploitation of player’s loss aversion is a socially
undesireable situation. The reduction of this exploitation
is recommended.

References
[1] Appfigures. 2018. War of Nations - PVP Strategy,
https://appfigures.com/reports/ ranking
[2] B.F. Skinner. 1959. Cumulative Record, Meredith
Corporation, Definitive Edition.
[3] D. Friedman, K. Pommerenke, R. Lukose, G. Milam, B. A.
Huberman. 2007. Searching for the sunk cost fallacy.
Economic Science Association, DOI 10.1007/s10683-006-
9134-0
[4] D. Kahneman, J. L. Knetsch, R. H. Thaler. 1991.
Anomalies The Endowment Effect, Loss Aversion, and
Status Quo Bias. Journal of Economic Perspectives,
Volume 5, Number1
[5] D. Draganov. 2014. Freemium Mobile Gaming - Design &
Monetization. CreateSpace Independent Publishing
Platform. ISBN: 1512322172
[6] G. Mangan, S. Sophister. 2016. Pricing's Next Top Model:
A Game Theoretic Analysis of 'Freemium' Pricing. The
Student Economic Review 2016.
https://www.tcd.ie/Economics/assets/pdf/SER/2016/Manga
n_Pricings_Next_Top_Model.pdf
[7] G. Spencer. 2016. Exploring the App Store’s Top Grossing
Chart. MacStories,
https://www.macstories.net/stories/exploring-the- app-
stores-top-grossing-chart
[8] J. Zagal, S. Björk, C. Lewis. 2013. Dark Patterns in the
Design of Games, Foundations of Digital Games
[9] N. Eyal. 2014. Hooked, Redline publishing company,
Volume 3, ISBN: 978-3-86881-536-8
[10] O. Clark. 2014. Games as a service - How free to play
design can make better games. Focal Press Taylor and
Francis group. ISBN: 978-0-415-73250-5
[11] T.Wijman. 2019. The global games market will generate
$152.1 billion in 2019 as the U.S. overtakes china as the
biggest market. https://newzoo.com/insights/articles/the-
global-games-market-will-generate-152-1-billion-in-2019-
as-the-u-s-overtakes-china-as-the-biggest-market
[12] “War of Nations“ Facebook group. 2018.
https://www.facebook.com/groups/195959667240038/
[13] “War of nations tips tricks and forum“ Facebook group.
2018.
https://www.facebook.com/groups/246277232203378/
[14] “War of Nations“ Facebook page. 2018.
https://www.facebook.com/warofnations/
[15] “War of Nations Forum“. 2018.
https://forums.popreach.com/forumdisplay.php?f=577
[16] W. Sims Bainbridge et al. 2008. Online Worlds:
Convergence of the Real and the Virtual. Springer-Verlag,
ISBN 978-1-84882-824-7

EAI Endorsed Transactions on

7 Context-aware Systems and Applications
Online First