Gold Farming Research Part Four

Analysis of Gold Farming from the Perspective of Industrial Sociology
The Historical Sociology of Gold Farming

One would hesitate to use the word “stages” in relation to a history of gold farming because the reality has been much messier. However, we can see some sort of chronology, framable in terms of Bernstein’s (1983) timeline of capitalist development:
• Subsistence production: in the beginning (1978¹), players produced and consumed items themselves.
• Barter: players producing a surplus of one item could barter with others who had a surplus of different items. This was, for example, part of Habitat, one of the first PC-based online games in the late 1980s (Damer 2007). As in real life, knowing (or at least trusting) other players could be an important component, as would be the existence of some form of trade contract enforcement. At this stage, the commoditisation of virtual items had begun – they had not only a “use value” (an intrinsic value based on their use or consumption in the game) but also an “exchange value” (a value in terms of other items for which they could be exchanged).²
• Monetisation: at some point (as noted in Section A1, Hunter (2006) tracks this to 1987), a trade jumped from barter of in-game items and currency, to exchange of real-world currency for in-game items. Thus virtual items (including in-game currency) began to be monetised. The commodification of virtual items was cemented as they began to have notional prices, as markets for their exchange started to arise, and as a division of labour emerged, with some players spending more of their time making items for others.
• Petty commodity production: at first, the production of virtual items/currency for sale would have been localised. We still see this today in the developing country players who sell at their own or nearby cybercafés (Magno 2006, Singh & Choo 2006). Likewise, players would have initially been selling only their own surplus items, but then gradually spending more time producing a surplus for sale, and less on just playing for fun. They thus started to specialise, moving on – again as we often see in developing countries today – from making money to pay for their games-playing to making cash for other real-world expenditure.

This brings us, then, to the situation around the turn of the century in MMORPGs. Virtual items were commodified to the extent that they could be exchanged, and players were starting to make money beyond that just required to play. Initially this might be called “gold market gardening” – an activity done alongside the player’s main occupation and alongside other gameplay. But the opportunity to earn led some to specialise in this, abandoning other gameplay and, in some cases, other occupations to take on virtually full-time gold farming. Thus gold farmers started to exist – mainly based in Western countries, and equivalent to an individual artisan or cottage industry model of production (see e.g. Ontain 2007).

From this point onwards, and seeing the money to be made, gold farmers looked for ways to expand their operations and cut their costs. They did this through three strategies; processes found in chronologies of capitalist development (Bernstein 1983) and the product life cycle (Porter 1980):
• Wage labour: initially gold farming operations expanded informally and based around real-world or in-game friends. At some point, though, gold farming took the leap from petty to capitalist commodity production when the first player paid another player to gold farm for them. In time, and as per the full capitalist commodity production model, some of the hired workers did not own the means of production: the PCs and software and even accounts were owned by the cybercafé owner or by the gold-farming firm owner. This is the model that pertains today in Asia, and has also been present in the West (see, for example, Concernedeq 2006).
• Globalisation/Offshoring: with wage labour forming by far the single largest cost of gold farming, it was inevitable that it would migrate to low-wage locations. Hence, the domination of Chinese playbourers in gold farming, which some see as having pushed US firms out of the market (Carless 2007).
• Automation: as seen in industries from automobile manufacture to software services, the key alternative cost-cutting strategy to offshoring is automation. Thus, rather than hire real workers, some gold farmers took to creating virtual ones (e.g. Kushner 2007). These are the “bots” that imitate the actions of real players, and which can be used to gather in-game items and currency.

The three strategies may be combined. Hence, the reports of Chinese playbourers minding multiple bots (Lee 2005). There may also be some geographical differentiation, with suggestions that American gold farmers have gone down the automation route, while wage labouring dominates Asian gold farms (Bell 2006, Mithra 2006, Seiler 2007). The two, though, may also be combined with US-based developers creating macros or finding bugs to allow duping, and then offshoring control of these to Asian workers (Lee 2005).

In charting the development of gold farming, then, we find that it fits quite well with standard industrial histories. Not just the Marxist analysis of commodification and capitalism described above, but also more business-oriented chronologies such as the international product life-cycle (Vernon 1996, Porter 1980) where innovation in an industrialised country is followed by rapid growth and intensifying competition, leading production to then shift to low-cost locations and markets to open up in those locations.

We can, therefore, use the frameworks of industrial sociology to help understand where gold farming has been. We might also use them to predict relations, underpinning factors, and where gold farming will go³.

Gold-Farming Value Chain Analysis

Analysis of the links from producer to consumer comes in various guises: global commodity chain analysis, filiere analysis, global production network analysis (e.g. Gereffi 1999, Raikes et al 2000, Henderson et al 2002). Here we will simply call it “value chain analysis”. That analysis identifies various stakeholders, arranged in a set of relations to one another, as shown in Figure 4.

Figure 4: The Gold-Farming Value Chain

To briefly describe, then, the stakeholders:
1. ICT suppliers: provide the hardware, software and telecommunications infrastructure necessary for gold farming to take place.
2. Gold farmers: those who work in-game. As the diagram indicates, they may sell direct to players, indirectly via brokers/exchanges, or they may work for a gold-farming firm.
3. Gold-farming firm: enterprises that undertake gold farming. As shown they may sell direct to players or indirectly via intermediaries. They will employ gold farmers and also workers undertaking support and other activities.
4. Brokers: probably not directly involved in gold farming; if not, they are classic middlemen who intermediate between producers and consumers by buying from the former and selling to the latter.
5. Exchanges: these put producers and consumers in touch with each other but, unlike brokers, do not directly participate in the trade.
6. Player-buyers: those who buy in-game currency, items and accounts, or avail themselves of power-levelling services.
7. Other players: “regular” game players who do not buy virtual goods and services.
8. Government and other institutions: provide the institutional framework within which gold farming takes place.

Two other actors are noted:
1. Other intermediaries: some gold farming firms will use intermediaries with payment accounts in Western markets to help trade with Western players (Jin 2006c).
2. Fansites: are online communities of players. Sometimes used to collectivise the views and actions of players, they are also used by brokers and sometimes firms for marketing.

We can understand the levels of power within the value chain from a resource-dependency perspective (Pfeffer & Salancik 1978). This understands the extent to which one set of actors is controlled by others in terms of the resources on which they depend.
• Individual gold farmers have little power. The resources they provide in skill, labour and in-game terms are readily substitutable by those further downstream in the value chain, yet they depend on those downstream actors for their key resource input: income. The substitution can occur either by switching to other gold farmers, or by drawing on the “reserve army of labour”.
• Gold-farming firms’ delivery of virtual items can be fairly readily substituted, but the resource of profile (and sometimes trust/reputation) they build cannot, so they have some power in the chain, though this is limited by their apparent lack of collective organisation.⁵
• Brokers and exchanges have more power through the resources of trust and reputation they build with other actors, particularly consumers. As discussed in Section B4, we can also see this as the intermediation power of information – they know of both farmers and consumers who, individually, could not “find” each other in the market without the broker. They can readily substitute both up- and down-stream actors, but are themselves rather harder to substitute. Brokers’ power increases if they buy up competitors and fansites.
• There are millions of player-buyers who are not collectively organised. They have little recourse against upstream actors: the law is unlikely to help; any complaints would be a speck lost in the sandstorm of criticism about gold farming; payment organisations like PayPal seem not to get involved. They are arguably readily substitutable, at least at an individual level, despite funding the entire value chain.
• Other players have had limited direct impact to date on the gold-farming chain; restricted to the occasional in-game farmer-killing sessions. This may change as at least one group has begun a collective legal action (Dibbell 2007). Their greater impact, though, is on the game companies through individual and collective complaints and, more important, threats to stop playing and hence stop paying. This provides an actual or threatened resource loss for the companies that they would take seriously.
• The game companies hold the greatest power including the “nuclear option” of closing down all game worlds, thus depriving all other value chain actors of resource access. Ultimately, everyone else is dependent on the game firms.

Of the other actors, one can say that government has more power than it has so far exercised. Governments could make it much harder for gold-farming to take place if they wished. Its continuation relies on the tacit tolerance or benign disregard of developing country governments. There has been some, limited, organisation among independent fansites, such as the creation of nogold.org which aims to organise commitment against gold farming, including blacklisting of gold farming web sites. However, many fansites rely on the advertising of gold-farming firms for their income.

If we look at how the distribution of power within the value chain has shifted over time, three things can probably be said. That power has become more distributed as more gold farming firms entered the sub-sector. That there was a potential power shift from brokers/exchanges to gold-farming firms as the latter set up their own online trading sites; but with that shift limited due to issues of trust. That there was an actual power shift to game companies as they began taking action against real-money trading.

Value chains are typically divided into those which are producer-driven (dominated for example by large transnational manufacturers as with cars) or buyer-driven (dominated by large retailers and marketers as with garments or toys) (Gereffi 1999). The gold farming value chain leans more towards the latter where the retailers and marketers would typically act as “strategic brokers”. But gold-farming is unusual in a number of ways. There is no dominant force given there is only one well-known broker (IGE) that no longer enjoys anything like a monopoly on trade So we cannot call this an intermediary-driven value chain and should perhaps call it a “competition-driven” or “market-driven” value chain. Yet the most powerful force of all – the game companies – are external to the main value chain.

We can also comment on two other features of value chains. First that “coordination of the entire chain [is] a key source of competitive advantage that requires using networks as a strategic asset” (Gereffi 1999:3). Brokers should be able to exercise this coordinating function and, indeed, it is what provides their power. However, the virtuality of the traded items, the anonymity of traders, and the immersion of both producers and consumers in an online broadband world mean this coordination function is at continuous risk of being undercut. Again, the gold-farming chain is unusual in this respect (though also arguably a portent of the future).

Second, that “organizational learning [is] one of the critical mechanisms by which firms try to improve or consolidate their positions within the chain.” (ibid.: 3). We know relatively little about competitive strategies in this value chain, though – as discussed in Section A2 – we have seen learning about new farming techniques and new delivery techniques to be one part of what gold-farming firms do. Another value chain competitive strategy is altering the nature of relations within the chain (Grantham & Kaplinsky 2005). Disintermediation of brokers by gold-farming firms has been one such strategy. Another has been the attempt of new brokerages and exchanges to set up on the basis of a collaborative rather than conflictual relationship with the game companies.

Lastly, we can consider value chain span (see Figure 5). In these terms, the gold-farming value chain can operate at various levels:

Figure 5: Spans of Different Gold-Farming Value Chains

• Local: players within a physical geographical location – either a single cybercafé or a cluster in one district – earn money by selling items and services to other players. This appears to be informal, with earnings being small (a few or a few tens of US dollars), irregular and used either to cover the costs of game play, or to provide a bit of extra “pocket-money”. Descriptions can be found from Malaysia (Singh & Choo 2006) and the Philippines (Magno 2006). This was also the origin of real-money trading in South Korea in 2000, with cybercafé (“PC bang”) owners enticing new gaming customers by allowing them to make real-money purchase of high-level items from experienced customers (Huhh 2008).
• National: players in a developing country buy from gold farmers located in that same country, often through country-specific portals. The most obvious example is China with gold farmers reporting that many of their buyers are Chinese (Honge 2006), and with active exchanges such as http://www.5173.com (or with gold and services sold on general portals like http://www.taobao.com). There also appears to be some type of national market emerging in India (Sengupta 2007).
• Regional: players in richer countries buy from gold farmers in a poorer country from the same region. As mentioned in Section A3, this could be seen as a form of “nearshoring”. It is most often reported in sales from China to players in Japan, Korea and Taiwan (Jin 2006, Jin 2006e), but there is also one example of a gold farm in Mexico selling to players in the US (Dibbell 2003).
• Global: players in one region of the world buying from gold farmers in another; of which the stereotype would be US players buying from Chinese gold farmers.

It is the global trade that has garnered almost all attention. However, it is entirely possible that the other three levels of trade are larger. For example, Table 1 noted the best guess for global sales (US$500m) was less than the two estimates for Korea alone. There are uncertainties about the basis for the Korean calculations but, at the very least, this supports a contention that the focus of gold farming research should be on all four levels, not just the global trade.

1: The first multi-user dungeon game – generally seen as the key forerunner of modern MMORPGs – was MUD written by Roy Trubshaw, an Essex University student. It began running in autumn 1978 (Bartle 1990). Multi-player gaming (though not with tradable items) long predates this. Damer (2007) and others cite Spacewar, running on a PDP-1 in 1962, as the first multi-(two-)player computer game. For a fuller timeline of early online games, see Koster (2002).
2: One other form of barter is reported from the 1980s in relation to power-levelling: “Men did it for women in the hope or expectation of some kind of emotional or physical relationship.” (Hunter 2006).
3: Though Marxists anticipating an overthrow of the bourgeoisie by the proletariat, as opposed to just the ongoing oppression of the latter by the former, may be disappointed. In the real world, gold farming labour is unorganised. The virtual world does see its riots and protests but it is uncertain they achieve much in structural terms.
4: Salyer (2007) claims the company had a 60% market share at the time (which would posit a US$180m total market in 2006).
5: In 2007, Korean real-money traders were reported to have banded together to form the “Digital Asset Distribution Promotion Association”, chaired by ItemBay’s CEO (Anderson 2007). However, this appears to have been a specific reaction to moves by the Game Industry Division of South Korea’s Ministry of Culture and Tourism to introduce a new bill prohibiting virtual currency (though not virtual item) trading (Paul 2006).

Related posts:

1. Gold Farming Research – Part One
2. Gold Farming Research – Part Two
3. Gold Farming Research – Part Three
4. Gold Farming Research Part Five
5. Gold Farming Research Part Six – the last one