Gold Farming Research Part Five

This is the section five of the series(gold farming research): Enterprise Analysis of Gold Farming! In this section, we turn to analyse gold farming using other frameworks from the business and management literature that fall within the ambit of enterprise analysis. We start by applying some classifications to entry, existence and progression in the sub-sector. Then competitive frameworks are used to analyse the enterprise context, discussing in detail the many threats that gold-farming firms seem to face.

So it has been sorted as following:

1. Analysing Entrepreneurs and Sub-Sector Entry
2. Classifying Existing Gold-Farming Enterprises
3. Charting Enterprise Progression
4. Competitive Analysis

Analysing Entrepreneurs and Sub-Sector Entry

Heeks (2008) – from Grindle et al (1989) and Mead (1994) – differentiates three different types of enterprise based on the motivations and context of the entrepreneur involved:

• Survivalists are those who have no choice but to take up the income-generating activity because they have no other source of livelihood. Income provided may be poverty-line or even sub-poverty-line. Most “entrepreneurs” in developing countries are of this type, and Mead describes them as “supply-driven”: forced into enterprise by push factors related to their poverty and lack of alternatives.
• Flyers are true entrepreneurs who have taken up enterprise because they see opportunities for growth. Income levels may meet more than basic needs, and enterprises may graduate to the medium-scale category. Only a very small proportion of developing country small entrepreneurs fall into this category, and Mead describes them as “demand-driven”: pulled into enterprise by factors such as the opportunity for profit.
• Trundlers fall in between the two other groupings and represent those whose enterprise turnover is roughly static and who show no great desire or no great capacity to expand. Income provided will be enough to meet basic needs. These form the second-largest group of small entrepreneurs in developing countries, and their stasis reflects the relative lack of strong external push/pull factors.

There is evidence of individuals seeking employment in the gold-farming sub-sector due to their unemployment and lack of other avenues (Jin 2006)^{1. There is also one note that a requirement in late 2003/early 2004 for Chinese cybercafés to close at midnight led some to start going out of business, and then look to gold farming as an alternative (PJ 2007). However, on the basis of available evidence, it would appear that most gold-farming firms are flyer-run; with at least some being set up by gamers and/or gold farmers who see an opportunity for higher earnings (Jin 2006f, Dibbell 2007, Wang 2008).}

Box 11: Alternative Typology of Gold-Farming Entrepreneurs
Shapero’s (1982) model – summarised below – explains why entrepreneurs move into enterprise.

Gold Farming Figure 6: Alternative Typology of Gold-Farming Entrepreneurs

Agboma (2007) uses this model to investigate entry into the information technology sector in a developing country. He concludes there are two types of entrepreneur:

• Technical entrepreneurs: who move into the field because of their interest in IT.
• Business entrepreneurs: who move into the field because they identify this as a sector with profit prospects.

A similar typology seems feasible within gold-farming: players vs. profiteers. The former – we may also call them “gamer entrepreneurs” – are attracted because of their interest in online games. The latter – we may also call them “business entrepreneurs” – are attracted by the profile and perceived profits associated with the sub-sector.

We can therefore surmise that most gold-farming entrepreneurs have been pulled into the market by awareness of demand coupled with availability of input factors. Those input factors would include Internet-connected PCs, game software, semi-skilled labour, and information on processes and consumers. Where they access another key input – their finance – is only reported in passing in a few case studies, but it appears to be mainly family and friends (Barboza 2005, Wang 2008)². Gold-farming enterprises therefore seem to show some traits typical of developing country enterprise: being mainly micro-enterprise-sized and informal sector-based (see below), often being set up by someone with prior experience in the sector, and often using informal finance sources (Heeks 2008).

The entrepreneur classification further helps to understand or predict a number of things on the basis of other evidence about flyers (ibid.). First, that such enterprises are unlikely to require much external intervention; instead, their main concern is that government and other agencies should “stay out of the way”. Second, that such enterprises have a disproportionate impact on developing country economies; for example in terms of contribution to employment growth, economic returns, and formation of medium-sized enterprises. Third, that such enterprises are likely to sustain longer than the other types but, if they do fail, the entrepreneur is most likely to set up a new enterprise.

Classifying Existing Gold-Farming Enterprises

Most of the literature focuses on cases of gold-farming enterprises with a few dozen up to some hundreds of staff³. However, Jin (2006b) states “The size and organizational structures of gold farms vary … they can be like large corporation, underground gang or home business”. Jin (2006) therefore reports finding hundreds of firms with “3 to 10 computers”, and Wang (2008) confirms that most gold-farming enterprises “have less than 10 people, a couple of computers.”

Definitions vary but it can be said that a modal definition of “large enterprise” is one with more than 250 employees, of “small enterprise” one with fewer then 50 employees, and of “micro-enterprise”, one with fewer than 10 employees (ILO 2008). Only three enterprises involved with gold-farming – IGE, Thsale and ItemBay – possibly fall into the large enterprise category; each reported to have around 300-400 staff (Bell 2006, Carless 2006, Salyer 2007). Instead, we can say that almost all gold-farming firms are small enterprises and, most likely, the great majority are micro-enterprises. We expect, then that the size profile will be similar to that found in other economic sectors; hence the use in Section A3 of 2.5 employees per average enterprise. However, that could be too conservative, with 6-7 employees per average enterprise probably fitting better with the two sources cited above.

It seems unlikely that this is a particularly concentrated sub-sector. Judging from the plethora of suppliers, it appears to be highly competitive and thus quite likely to be dispersed. The web traffic data from Appendix 3 seems to bear this out, though one game company claims there is collaboration between leading gold-farming players (Seiler 2007).

Running through some other basic enterprise classification schema (Heeks 2008), we can say that gold-farming enterprises are:

1. Services not goods producers; though they share some characteristics of goods producers because they trade in virtual goods.
2. Consumer not producer enterprises: they sell to end consumers not to enterprises which will use their outputs for further processing.
3. Both domestic- and export-market oriented.
4. Privately-owned.
5. A mix of registered and unregistered. Jin (2006) reports some gold-farming firms in China being both registered and paying tax, but others as sitting within the informal sector: unregistered and not paying tax.
6. Mainly male-managed. Such evidence as there is (e.g. Lee 2005, Jin 2006) does not mention female managers.

We have no specific data on enterprise age, so can only note generic figures for small enterprise in developing countries: that 15-25% of any cohort will have started up in the previous year (Heeks 2008).

Charting Enterprise Progression

Having classified some of the static features of gold-farming enterprises, we now turn to dynamics. We can view the issue of progression in two related but so far unanswered ways. First, charting how a gold-farming enterprise progresses: “does it go anywhere?”. Second, charting how a gold-farming entrepreneur progresses: “can developing country entrepreneurs use it as a stepping stone to something else?”.

In terms of the former question, we could measure progress in the standard ways such as employment or turnover or profit. We appear to have no dynamic data to help on this. Or we could measure in terms of enterprise lifespan. The only data here is Jin’s (2006c) passing report on an entrepreneur closing his firm due to competition and inability to access Western trading channels. With no other data, we again note generic developing-country small enterprise figures: that 50% would be gone within three years, and 75% within ten years (Heeks 2008).

In terms of both questions, we could frame – even if not provide – an answer by invoking the idea of technological capability. Following Lall (1987), one may define technological capability as the general ability to undertake the broad range of tasks outlined in Table 6, and technological development of a gold-farming enterprise as growth in the capability as defined by movement up the categories and regardless of whether or not the final stage is attained. These capabilities are actually embodied in the skills and experience of individual workers, often seen as the most critical resource for ICT-based industries (Kumar 1988). In this case, technological development will particularly represent the accumulation of increasingly skilled workers.

Level1. Non-production operational capabilities

Playing a software game

Choosing a software game

Training others to use the game

Playing all aspects of the game

Installing the game

Troubleshooting the game

Adaptation without production

Creating a situation-specific application within the game; e.g. by using macros

Hosting a game

Simple game production

Creating a very basic game with a game production package; e.g. GameMaker

Creating a small utility program or modification; e.g. a bot

Intermediate game production

Creating a basic game; e.g. a simple mobile game

Creating one element of a game; e.g. artwork

Recoding language and interfaces to meet local user needs

Redesigning a game program to meet local user needs

Redesigning a program to meet regional/global user needs

Minor process change: modifying the game production process

Skilled game production

Local product innovation: developing a new game to meet local user needs

International product innovation: developing a new game to meet regional/global user needs

Major process change: redesigning the game production process

Process innovation: designing a completely new game production process

All gold-farming firms are, of course, operating at at least Level 1 and almost certainly Level 2. Some firms could well have developed Level 3/4 capabilities, since – as discussed earlier – they are reported using macros, bots and other game modifications, and they do have research capabilities. In capability terms, it would be a step further forward to move to more “mainstream” game industry activity. Such activity certainly exists in the same developing countries as gold farmers. What we do not yet know is whether there is any progression or connection between the two, as we have only the following scraps of data:

• Progression along the value chain (i.e. moving from a production role to a sales, marketing and production role; not the same as technological progression) is seen as being constrained by lack of skills (particularly English) and contacts (particularly in Western markets) (Jin 2006e).
• One gold-farming entrepreneur mentions his wish to progress by staying in the IT sector but graduating on to work in e-commerce (Barboza 2005).
• One gold-farming manager describes his personal progression into games design (PJ 2007).

If progression up the technological capability scale were possible, it would certainly be valuable. As discussed next, gold farming is a very precarious way to earn a living. Higher-skilled work via the games industry would be less vulnerable and also better paid.

Competitive Analysis of Gold Farming

There are various ways in which one could approach competitive analysis of gold-farming. For example, one could do a simple SWOT analysis, mapping the current internal strengths and weaknesses of gold-farming firms, and future opportunities and threats. Strengths could relate to playing skills, low costs and ready supply of labour. Weaknesses could relate to poor management and marketing, staff turnover, and – for some probably – questionable ethics. Opportunities would include the demand caused by growing numbers of online players worldwide, with particular growth in Asian markets including growth spurred by the emergence of an Asian games industry⁴.

One could use Porter’s (1990) “diamond” (see Figure 6) to analyse how China has been able to create a competitive advantage in gold-farming48; most likely thanks to elements such as:

• Factors conditions: strong ICT infrastructure, and available or trainable skills.
• Demand conditions: strong demand growth both nationally and overseas.
• Related and supporting industries: the growth of both the cybercafé and games industries within China.
• Firm strategy, structure and rivalry: the drive to enterprise in China since the 1990s, and strong competition between firms.
• Other: government’s mix of supportive intervention on infrastructure and non-intervention on issues like legality/formality, and the chance of proximity to Korea.

Gold Farming Figure 7: The System of Competitive Advantage

Or, one could use Porter’s (1980) “five forces” model (see Figure 7) to understand the nature of competition in the gold-farming sub-sector. This would again cover the high level of competition between gold farms (reflected – see Section B2 – in the declining price of virtual goods and services). It would identify the relatively low barriers to entry: economies of scale are relatively limited for producers, as discussed in Section B3; firms are poorly differentiated; capital requirements are low; and customer switching costs are low. Only the issue of access to distribution channels is a potential entry barrier. One would draw on value chain-type analysis to say that gold farming firms’ suppliers have limited power because switching costs are low and supply is relatively competitive. Buyers would be analysed similarly to argue that brokers and exchanges have more power than individual player-consumers.

Gold Farming Figure 8: Competitive Forces Framework

But in all three cases, we so far have left out one element – threat. As plain “threats” in the SWOT model; as an element in the analysis of demand and government and rivalry in the diamond model; and in understanding the power of buyers and the threat of substitution in the five forces model. We analyse this element here in much greater detail because risk and uncertainty and vulnerability seem so strongly to characterise the gold-farming sub-sector.

Other than competition/new entrants, an obvious threat is rising labour costs, an issue across all Chinese sectors and in Asia more broadly (Reuters 2008, UN 2008). As mentioned in Box 9, government action may also be a threat. Western governments so far seem more interested in the taxation of virtual income than in gold farming (Crooks 2006). Governments in Asia have been more engaged, with the Japanese government launching an investigation into real-money trading (Wyman 2006)⁶, and with the Korean government enacting legislation against it (see Box 13).⁷

However, the main focus in the remainder of this section will be the threat from other players and, more, from the game companies. The discussion that follows looks first at identification of gold farming; then at the responses to gold farming, and at the incentives to respond; and finally at the impact of those responses.

Identifying Gold Farming

Following the discussion in the gold farming research part two, we can see gold-farming firms tend to have three main types of in-game presence:

• Low-level grunts: characters that advertise the services and web site of the firm to other players, or which mule money and items around. – The thing has been changed, such as the wow gold farmer are smart to understand that it is able to grind wow gold effectively if they use the high leveling character to lead the lower.
• High-level earners: who have in-game powers that enable them to either get high-level (i.e. high-worth) items and/or earn significant amounts of in-game currency through their avatar.
• Bots: automated avatars that undertake repetitive money-making activities (mining ore, chopping wood, killing gremlins, buying from NPCs), or repetitive activities such as advertising. It’s really, they are many bots in the popular game such as world of Warcraft. Especially in Aion Online, We can’t imagine that there are too many bots in a new game published in 2009. So you should learn that how to choose a solid wow gold store.

Gold farmers are also present in-game when they take over a player’s character for power-levelling.
Avatars advertising gold-farming services are easy to recognise. Otherwise, in-game recognition of gold-farming avatars relies on rather unscientifically putting together clues (Paul 2005):

• Behaviour: e.g. trying to remove other players competing for high-level resources or items; constantly doing the same thing or being found in the same gameworld area.
• Appearance: e.g. low-level avatars may be dressed in the standard clothing or equipment that all new players are issued; characters may be in play 24/7.
• Level: e.g. either level-1 or a high level.
• Interaction: e.g. do not answer interactions, or answer in poor English. Other than actual advertising this, above all else, seems to be regarded as an acid test: “English speaking gold farmers tend to go unnoticed because they can communicate with players (and so say they aren’t farming, or move to a different farming area if players complain) and because if you can respond you’re generally not seen as a “chinese gold farmer”.” (GrinR 2005).

Game companies have some additional sources of recognition evidence. These vary from game to game and company to company, though they still rely somewhat on perception. They may include:

• Out-game monitoring: gleaning information from auction sites and gold farming sites and trying to link to in-game characters.
• Circum-game monitoring: tracking the IP address⁸ of log-ins to see those that suddenly jump from one location to another (found with power-levelling); tracking how long a character is playing (may be 24/7 with power-levelling); looking for use of automated code (bots/macros).⁹
• In-game monitoring: of messaging (for advertising); of trading (large amounts of cash or high-cost items being exchanged for nothing between players).

Responses: The Threats to Gold Farming

If they are recognised, gold farmers face threats on a rising scale:

• Abuse: rude messages from other players. A number of gold farmers report being the subject of abusive in-game messages (e.g. Jin 2006e).
• Reporting: other players can report suspicious or prohibited behaviours to the game company. This only causes a problem if it leads to more serious repercussions. Many players will have done this in practice. Over time – for example with World of Warcraft in 2007 – game companies have made it easier to report spamming and other RMT-related actions. They also rely on reporting to guide further actions such as patching and account banning (e.g. Square Enix 2008).
• Bot disruption: bots work by repeating certain program code; some game companies try to combat this with “random events”: unexpected activities that only a human can resolve and which will stop the bot from working. In Runescape, for example, a tree being chopped may turn into an Ent as a random event. If the avatar keeps chopping, their axe will break. Regular players may also attempt to disrupt perceived bots.
• Character attacks and killing: where allowed, other players may kill avatars they suspect of gold-farming or may induce non-player characters to kill those avatars. To see this in practice, type “killing gold farmers” into YouTube or Google and peruse the results (also Weir 2004, Dibbell 2007). Gold farmers themselves also describe experiences of other players deliberately killing them (Bell 2006).

Random events may include attack from a non-player character; an example in Runescape is the arrival of an evil chicken which will kill players that keep doing whatever they are doing and fail to run away or fight back. Likewise Square Enix installed a high-level monster in a Final Fantasy XI game area where low-level gold farmers were seen to be fishing a lot, in order to kill or deter those characters. In most games, avatar death is only a temporary disruption but may lead to some item loss.

• Nerfing: where gold farmers are found to be making use of some particular item or aspect of the game to make money or to power-level, those items/aspects may be “nerfed” – downgraded or even potentially removed from the game by game company programmers.
• Account banning: where – through monitoring or reporting – game companies believe they have identified gold-farmer characters, those characters and their associated accounts may be banned. Some subscription fees may be lost. It takes time and effort to resubscribe. And many hours or days or even weeks of work may be lost if a high-level character was banned. In May 2006, to take just one example, Blizzard banned more than 30,000 accounts on World of Warcraft, citing automated gold/item farming as a key reason behind the bans (Reimer 2006); ArenaNet claims to ban “typically more than 5,000″ accounts per week on GuildWars for real-money trading (ArenaNet 2008); and Jagex claims to have removed more than US$2.6m-worth of in-game currency by account banning during 2007. Gold farmers also report this: having customer accounts banned when power levelling (Carless 2007), and having their own gold-farming accounts banned and having to start again (e.g. on a daily basis: Davis 2007b).
• Other patching: a number of the changes described above are introduced into an existing game by the distribution of “patches”: sections of programming code that replace earlier sections. Other patches may seek to alter the trading system to prevent RMT-related actions. For instance, in 2006 and 2007, patches were introduced to both Knight Online and World of Warcraft which delayed delivery of payments when trading, making it easier to identify and block prohibited trades.
• Game redesign: the boundary between patching and redesign is fuzzy, but some game companies undertake quite fundamental changes to their games to seek to get rid of gold farmers. One example, described in more detail below, was the major change introduced to Runescape by Jagex in 2007.

Beyond these, further threats include:

• IP banning: companies may block logins from certain IP addresses they believe have been associated with prohibited actions. Such things can be worked around, but with some difficulty. For example, Blizzard bans all Chinese IP addresses from its North American and European servers (Zhe 2006).
• Blocking channels: shutting down particular sales or marketing channels used by gold farmers. In 2002, Mythic Entertainment (makers of Dark Age of Camelot) had all eBay listings for a gold-farming firm – Black Snow Interactive – removed (Dibbell 2003). This approach then grew such that, by 2007, eBay was delisting all gold farming auctions, reportedly because of pressure from game companies (Dibbell 2007). As another example, in 2005, PC Gamer magazine stopped carrying RMT adverts, reportedly because of pressure from gamers (Carless 2006).
• Legal action: game companies or others may file lawsuits against gold farmers if they believe a case is provable. In the most serious case, the gold-farming firm would be put out of business. In 2008, Blizzard (makers of World of Warcraft) successfully won a legal injunction against In Game Dollar LLC (apparently run by a Florida-based individual) which operated the peons4hire web site. The web site was closed down (Duranske 2008). At the time of writing a one-year class action law suit against broker IGE was working its way through the Florida courts (Virtually Blind 2008).

The Incentive to Respond on Gold Farming

The incentives for game companies to take action on gold farming are not simple. On the positive side of the account, for pay-to-play games, companies earn revenue from gold farmer subscriptions, and other players may be more likely to play or keep playing (i.e. keep paying subscriptions) if they are able to buy currency, items and high-level characters¹¹. Put economically, gold farming is utility-maximising for both parties – gold farmer and player-buyer – otherwise, of course, it would not take place (Castronova 2002). Doing nothing about gold farming also costs nothing whereas doing something costs money in staff time and other resources.

Doing something may also produce unforeseen costs, as can be noted in the experience of Jagex. In 2007, in response to continuous complaints from players of Runescape about botting and gold farming, Jagex introduced a series of game changes that made these practices much more difficult. Bots do seem to have largely disappeared as a result. However, gold farming still continues. Methods have changed but currency exchange rates appear to be near their historic levels. All this cost Jagex a significant amount of money. Yet it has led many players to complain that the nature of game-play is now seriously degraded; causing them to rename the game Ruinedscape. Numbers logged as playing the game now appear lower, though this may be accounted for by removal of bots. Only time will tell whether this high-cost, high-risk strategy against real-money trading has worked. It does suggest that regular players may be very ready to complain about gold farming, but less ready to bear any personal cost for its removal; and that gold-farming actions may create unexpected disutilities.

For pay-to-play games, there is only one main item on the negative side for “legitimate” gold farming: when gold farmers’ actions or the perceptions of those actions are such that players cancel their subscriptions¹². In economic terms, again, gold-farming may increase the utility of producer and consumer but also have externalities that reduce the utility of a third party: other players (Castronova 2006) (see also Box 4). It is not clear that much has been researched and written on this specific issue: the actual economics, and the size and motivation of any gold-farming-related cancellations. However, reports from game companies (e.g. Ward 2008) provide some insight. Loss of “immersion” is reported as a problem: it is hard to keep imagining you are a medieval knight defending your kingdom if you’re being bombarded by exhortations to visit various web site addresses. The perceived presence of bots or macroing in-game or of gold farmers monopolising a particular area (“spawn-hogging” or “spawn-camping”) are also seen to damage immersion or game-play. Finally, players also complain about “kill-stealing”: gold farmers monopolising a monster they were trying to kill, and “ninja-looting”: gold farmers taking some resource or item while other players are occupied fighting the monsters that guard it (Steinkuehler 2006).

Reports on account banning in Final Fantasy XI, suggest 46% were for actual farming (building up currency in-game for later sale); 40% were for use of macros; 6% were for hogging/camping; 5% were for actual real-money trading; and 3% were of resubscriptions by players previously-banned for RMT (Square Enix 2007). Reports from gold farmers do include admissions of spawn-camping, kill-stealing and killing regular players to keep them away in order to meet quotas (Jin 2006e, Yee 2006).

Beyond specific in-game perceived experience of gold farming, players also have out-game sensibilities; either that it is “not fair” that some players should use gold farming to shortcut the time-consuming, arduous road to wealth and high levels; or less explicitly because the shortcuts highlight just how much time they themselves have been investing/wasting¹³. Gold farmers are also blamed by players for various in-game effects – inflation, deflation, oversupply of items (see Box 5). These effects seem largely seen as negative because they reduce either the purchasing or earning power of players. Kaminski (2006) reports that players divide approximately into thirds – one third don’t object to gold farming and may participate in it; one third don’t care; and one third are against it.¹⁴ Of the latter, more seem concerned about the perception of fairness/cheating than perceived damage to the game.¹⁵

Given the core role played by perception – “The real problem is the perception of gold farming, not the fact of gold farming” (Davis 2007d) – the incentives may be more for the game companies to be seen to be doing something than to actually be doing something. They do, for example, seem less concerned about covert than overt gold farming activities (Kazemi 2007). Despite this, it seems unlikely that the many reports of account banning are shams; and the legal actions certainly are not. Either the cost-benefit balance of taking action swings in the latter’s favour, or game companies are motivated by more than simple financials. Given most company staff are also players, they may also be swayed by some of the “moral” or immersion arguments about game integrity.

In summary, online game companies seem to have started life with an “Ignore It” strategy which changed at some point – particularly during 2005-2007 – to the “Fight It” strategy of patches, bans and legal actions described above. Alongside these two are three other strategies that game companies have adopted:

• Permit It: this has involved game companies formally supporting the creation of third-party player-to-player real-money trading portals. Examples include PlaySpan (set up in the US by non-resident Indians and with offices in India, China and Korea) and, forthcoming at the time of writing, RealityBridge.
• Host It: instead of outsourcing the trading portal, the game company hosts it. This was the approach taken by Sony Online Entertainment, which set up “Station Exchange” as a trading mechanism for two of the servers hosting their game Everquest 2, taking a 10% commission and making more than US$250,000 in its first year (Robischon 2007).
• Become It: though relatively little known in Western online gaming, Asian gaming has made significant use of the free-play, item-pay model in which gamers pay little or nothing to join the online game, but then pay real-money (typically deducted from pre-paid cards) to buy items in-game (e.g. ZT Online – Egan 2008). Other games provide their own power-levelling services. For example, Origin, who operate Ultima Online, sell level-85 characters ready-made for US$29.99. In the case of either selling items or selling high-level accounts, the game companies are earning money from exactly what the gold farmers do.

For developing country gold farmers, the Ignore It strategy will work best. Permit It/Host It strategy impacts will depend on what is and is not permitted. As discussed next, the impact of Fight It is debatable. Become It looks, at least on the surface, to be the most threatening – it takes us back to Figure 7′s five forces model, with the threat that gold farmers’ activities will be entirely substituted by the game companies.

The Impact of Game Company Actions

Game companies claim that their actions reduce the extent of gold farming and other RMT-related activities (e.g. Zonk 2007). There is reasonable data on what game companies say they are doing about gold farming. There is less about what they are actually doing and why they are doing it. And there is even less about the impact of their actions on gold farmers. So we know about the threat in theory, but very little about the threat in practice.

There are a few tidbits we can cite:

1. As mentioned already, gold farmers do report having their accounts banned, and having to start again with low-level accounts (Davis 2007b, Dibbell 2007).
2. One Chinese manager reports losing thousands of dollars-worth of virtual stock each time he had a “close-down” by the game company (of which he had had five) (Lim 2007).
3. In-game experience and reports from other players do indicate fewer bots, less spawn-camping and kill-stealing after anti-RMT campaigns (Orii 2007).
4. Out-game one can see web sites that have closed down (e.g. peons4hire.com), or the removal of certain games (e.g. Runescape during 2008) from gold-farming portals¹⁶

However, against these morsels we have two opposing items, albeit of less direct evidence:

1. Currency rates: as discussed in Section B2, in-game currencies have deflated fairly steadily against real-world currencies since the mid-2000s. All other things being equal, one would have expected effective anti-gold farmer actions to reduce supply and thus drive the currency price up. That this has not happened suggests gold farmer numbers/supply may not have greatly decreased as a result of game company actions.
2. Business processes: gold farmers have constantly proven themselves to be highly innovative. Not just in exploiting emergent game loopholes and bugs but in metamorphosis: “rolling with the punches” and changing their processes as conditions change. For example, when Blizzard banned non-US IP addresses from its US game servers, Chinese firms rented US proxy servers and got contacts in the US to pay subscriptions via their credit cards so they could continue to have access (Zhe 2006). When Blizzard began focusing on player-to-player trades, gold farmers started listing cheap items at very high prices in World of Warcraft’s auction system as a way to move money between characters (Torres 2007). Since one-sided trading of large quantities of currency between players became impossible in Runescape in 2007, gold farmers now take over the player’s character and provide the gold in that way (e.g. PCGamerUSA 2008). As with virus writers and anti-virus software, then, the sense is that gold farmers find ways to work around most roadblocks and that the game of cat-and-mouse will continue forever.

From the limited available data, it seems that gold farmers face a series of ongoing threats of varying severity. Understanding, avoiding, and mitigating these threats are probably a central part of daily activity for gold farmers. Porter’s models above encompass some aspects – threats to demand, threats by government, threats of game company substitution. However, in general, those models do not easily find a place to classify the majority of threats; most likely because the models are intended for mainstream industries not the type of “shadow industry” that gold farming represents.

Although the threats are real and do demand reactions from gold-farming firms, the level of this reaction is not entirely certain. Other than legal action, it may be that the impact is more at the level of nuisance and accepted cost rather than fatal intrusion into the lives of gold-farming firms. Nonetheless, threats will help to shape the competitive strategy of those firms – shaping the research they seek to do; shaping their choice of low-cost vs. niche vs. differentiation strategy (most currently appear to choose the former); shaping their reinvestments; and, ultimately, shaping their continuation in the sub-sector.

1:In more detail, the motivations of those who become gold farmers are reported as a mix of the push of unemployment, underemployment or poorly-paid work as the only alternatives; and the pull of those who are already gamers finding a way to make money from play (Jin 2006e).

2:There is one reported example of a US-based entrepreneur investing in Indonesian and Chinese gold farms (Lee 2005), and one imagines that Asian emigrants based in the West would have invested. As noted above, Chinese local governments are also reported to have made some investments (Jin 2006).

3:Staffing sizes mentioned in case enterprises are: 10 (Dibbell 2007), 20 (Wang 2008), 30 (Johnson 2006), 40 (Carless 2007), over 40 (Honge 2006), 50 (Dibbell 2007), 80 (Jin 2006c), 100 (Lee 2005), 300 (Salyer 2007), 400 (Bell 2006), 500 (Zonk 2007).

4:Not just the existing strength of the Korean games industry and the recent emergence of China, but also MMORPG producers or co-producers/localisers materialising in countries like VietNam (Phong 2007) and India (Sify 2008)

5:Or one could turn this inside-out to argue the competitive fragility of gold-farming based on poor skills and strategies, uncertain demand, destructive competition, etc.

6:Though this appears to have resulted in a laissez faire policy outcome (Davis 2006).

7:An indirect impact on gold farming might come from generic game-related concerns of governments in Asia. These seem to be about the level of violence depicted and about the amount of time young players are investing in gameplay (Wahid et al 2006). On the latter, the Chinese government, for example, has enacted restrictions both out-game (e.g. limiting cybercafé hours) and in-game (e.g. halving game points for under-18 players (all players must register with their ID card details) after three hours, and reducing to zero after five hours) (China Daily 2007, PJ 2007).

8:The Internet Protocol address – the unique four-number code that identifies each Internet-connected device.

9:Game companies have access to an immense amount of game data. They are thus able to use data mining to identify issues for further investigation (Kazemi 2007). For example, plotting data on all trades as a network, or all quest completion times to identify anomalies. Other game company tactics include use of monitoring software to detect use of bots and other cheats, such as Blizzard’s Warden system.

10:There is a political hinterland to this legislation. In an effort to prop up declining levels of business, Internet cafes and video arcades began running video “slot machines” that paid out not money but “tickets” that could be used for government cultural events and site visits (Koster 2006). However, the machines – particularly those running a game called “Sea Story” – were hacked to increase their payouts. The tickets also came to be used as an alternate currency that could be traded for real money. As a result, the “Sea Story” machines became incredibly popular; earning a reported US$15bn per year for the café/arcade owners. With such vast earnings, corruption of public officials (politicians, bureaucrats and ministers for licences and/or to influence legislation), tax evasion, and reports of money laundering and the involvement of organised crime gangs, soon followed. The ban on virtual currency trading was one part of the government’s reaction to the ensuing scandal and public pressure. MMORPGs, therefore, may not have been the target of the legislation and it is, at the time of writing, unclear to what extent they will in practice be caught in the crossfire.

11:This is implicit within the comments of some players who have bought gold-farming services (Yee 2004), and more explicit in the explanations of players who cancelled subscriptions. Many did so because of the difficulty and/or dullness of raising their character’s levels (Yee 2003).

12:Castronova (2006) hypothesises that gold farming also adds to company costs by increasing the cost of customer services due to the number of gold-farming-related complaints that will be lodged. That does seem probable: Robischon (2007) reports that virtual item sale disputes took up 40% of Sony.

13:One interesting counter-argument is that all players are putting time and effort in; regular players invest that direct in the virtual world; players who trade with gold farmers are investing that in their real-world jobs, for which they are paid, and then transferring their pay as an investment into the virtual world.

14:Kaminski’s player sample (n=26) is small but these figures chime exactly with those of Castronova (2006) drawn from discussions with games industry managers who cite one-third of players participating in RMT, one-third against it, and one-third falling into neither camp.

15:For other player testimony see, for example, “The Chinese farmers have utterly ruined the economy and unbalanced any sense of fairness in the game.” (He 2005); also Dibbell (2003), Yee (2004), Bell (2006), Yee (2006), Orii (2007) plus NoGold (2008).

16:There have also been declines in the level of Google searches for some in-game currency since the mid-2000s. For example, see Google Trends for “wow gold” or “ffxi gil”; though what this indicates is ambiguous.

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 Four
5. Gold Farming Research Part Six – the last one