Some Edinburgh City councillors already privately refer to the city’s tram project as the problem that “cannot be named”. Much as actors refer to Shakespeare’s tragedy as “the Scottish play”, superstitions of bad luck now bedevil the production. A dramatic shift from the optimism that initially characterised the development of the Edinburgh tram, towards pessimism.

That which cannot be named is no longer just the failure of a flagship local transport policy. The issue has engulfed the City of Edinburgh Council, and now risks destroying local politics completely: Not only the existing administration, but public trust in local government decision-making.

Political heavy-weights, who normally shy away from the minutiae of local governance, are now offering parental guidance in public: Alistair Darling (local Member of Parliament, and former United Kingdom Chancellor and Secretary of State for Transport) described the option to borrow £231 million ($370 million) to complete the city centre section of the tram line as “absolute madness” – the local population would be saddled with vast debts. Days later, Graham Birse (chief executive of the influential Edinburgh Chamber of Commerce) called the decision to not complete the city centre section, “bonkers” – far fewer passengers would use a tram that did not serve the city centre adequately. Even Alex Salmond (Scotland’s First Minister) has become directly embroiled, struggling to contain calls for an immediate public inquiry to identify who is responsible.

Burn the witches! This Scottish tragedy is rapidly descending into farce. That would be unfortunate, because this particular local difficulty goes to the heart of the Scottish nationalist agenda: A desire for greater devolution of public funds to local level. More localised independent entities have fewer financial resources, so are less able to manage expensive, risky projects. Consequently policy ambitions also need to be scaled back. Such scale isn’t necessarily a problem – small can be beautiful. The problem lies in pretending to be big, when not.

This article introduces the concept of risk in tram (and similarly large public transportation and infrastructure) projects, chronicles the decisions that lead a relatively small local authority to need to find hundreds of millions of pounds to support a single project, and explores the implications for future policy-making, especially in the context of a more devolved Scotland. On this page:

• Tram Cost Uncertainty
• Devolution of Chaos
• Scaling Ambition
• Appendix: Policy History of Edinburgh Trams

Tram Cost Uncertainty

I have previously discussed why estimations of the cost of large transportation infrastructure projects (especially light rail) are both inherently inaccurate, and tend towards optimism. Edinburgh Trams transpire to be an extreme example: From £375 million estimated for the original 3-line network, via £545 million for a more pragmatic 2-line network, to over £1 billion for today’s single line (which is still far from completion).

However, this uncertainty was somewhat predictable: The graph below is taken from Bent Flyvbjerg’s 2004 Procedures for Dealing with Optimism Bias in Transport Planning, guidance issued to central government alongside HM Treasury’s Green Book. It shows the probability that the cost of rail projects (including trams) will exceed the estimated budget.

Distribution of Rail Cost Overruns (Bent Flyvbjerg, 2004)

(Data is based on analysis of 46 rail projects from across Europe and North America – UK has few recent rail schemes, however comparison of road schemes suggests patterns are very similar for all regions.)

The average cost overrun is about 40%. In the planning/appraisal process 40% is literally added to the estimated cost of a project as “optimism bias”. Optimism bias is part of Scottish Transport Appraisal Guidance, to which the Edinburgh Tram was subjected, although the original tram analysis pre-dates formalised optimism bias.

The 40% value should protect the UK treasury:

• Over many separate projects, cost overruns will average to zero. Broadly, in the long-term, the treasury will remain balanced, without requiring individual projects to be micro-managed from the top of government.
• Individual project costs are in the hundreds of millions (£). These are still a tiny proportion of UK Gross Domestic Product, taxes collected, or ability for the UK government to take loans. The national economy will not be thrown into a crisis if one specific project goes badly wrong.

Unfortunately, merely adding 40% adds a sense of certainty to an individual project which remains uncertain. For the immediate funders of a project, the important question is, can we fund a cost overrun of 80%? For the City of Edinburgh Council, the answer to that question was effectively, “we cannot”.

A quick glance at the council’s budget puts the tram in perspective: The council’s total annual capital budget (for investment in everything) is only £235 million, less than a quarter of the capital cost of one tram line. Typically transport accounts for 20% of local government capital expenditure, so we might expect a local authority like Edinburgh to be investing around £50 million each year in transport projects. Placing over 10 or 20 years worth of investment into just one project suggests a tram scheme was far too ambitious to ever be a local government responsibility.

Not only do councillors appear to be losing a high-stakes casino game, but they seem to be playing with all theirs chips on the table.

However, the tragedy is not so simple, because while the City of Edinburgh Council are responsible for the tram project, they are not the only funder:

Devolution of Chaos

Conventionally in Britain, higher tiers of government act as financial guarantors for low tiers. A local authority has statutory (in law) responsibilities, and hence cannot “go bankrupt”, however badly it manages its budget. This hierarchical structure is not accidental: It is rooted in currency (money itself), for which central government is solely responsible. In practice central government imposes strict financial controls on local government, which limit the scope for mis-management. Historically important cities, such as Edinburgh, also tend to own a lot of local assets (like property), which can be sold in a crisis.

1990s Scottish devolution did not devolve money, it just altered the hierarchy slightly, with an extra decision-making tier (the Scottish Parliament and their civil service, the Scottish Executive) in between Edinburgh’s council and the United Kingdom’s central government.

In addition to keeping a watchful eye on how the council manages its budgets, the Scottish Parliament is part-funding the Edinburgh Tram – providing extra money beyond normal budgets or spending approvals. The Scottish Parliament’s contribution (currently via Transport Scotland) is limited to £500 million [PDF]. And herein lies the problem: When that funding package was agreed, the local council’s contribution was the remainder, just £45 million, a value broadly attainable by the local authority. But since Scotland’s contribution is fixed, every time the estimated price rises, the City of Edinburgh Council’s contribution rises out of proportion: Without other sources of funding, a doubling of project cost to £1 billion is actually a ten-folded increase in the cost to the council…

By attempting limit the Scottish Parliament’s exposure to the tram project, all the risk has been transferred down the hierarchy, towards the tier of government least able to raise large amounts of cash in a crisis.

Brinkmanship of the worst kind? In the interim, the council’s response is to control escalating costs by reducing the length of the route to be completed, with apparent disregard for whether the resulting tram track connects places large numbers of passengers might want to travel between. Underpinning their decision appears to be one of the worst assumptions of “modern” accountancy: That the asset value of a (tram) service is defined by the cost of constructing the infrastructure on which that service operates. Strategically, the value of the (tram) service to the city of Edinburgh is linked to the number of passengers that use it. Cynically, to the number that see it. Even commercial assets that are technically transferable (like land and tram vehicles) tend to sell for a lot less than they were bought. Scottish government’s reaction to the council’s decision was to withhold part of Scotland’s funding contribution, effectively forcing the council to reach a different decision [the following Friday, 2 September, the city centre section was added again].

With Scottish national government substantially funding the Edinburgh tram, and the project appearing too risky for local government, perhaps the Scottish level of government should have conceived and manage the project from the outset? The appendix at the bottom, Policy History of Edinburgh Trams, will help explain.

Haymarket Tram Terminus? Existing bus to Edinburgh Airport, with new tram station “under construction” on the left (August 2011)

Scaling Ambition

Broadly, there are 2 viable methods of containing risks on the scale of the Edinburgh tram:

• As a Scotland-level public project: Even £1 billion is only around 1% of Scotland’s Gross Domestic Product, a magnitude of (over-) spending that should be manageable by a truly devolved Scottish government. The Scottish Parliament has subsequently established an organisation (theoretically) capable of managing transport projects at this level, Transport Scotland.
• As a primarily private project: Most modern tram systems built in England transferred risk to the private sector (rather than Edinburgh’s model of merely contracting private sector suppliers). Scotland’s past record on private sector transport projects is poor (most obviously on the Skye Bridge), because policy changes too fast.

Unfortunately neither of these methods complement good local public governance: Decision-making and responsibility becomes remote from local politicians. Ultimately, this would dilute the role of local authorities: Logically to organisations that deliver (statutory) local services and act as “community council” talking shops, with no ability to actively shape or invest in the long-term future of their own areas.

That may be where Scotland is heading: Far greater centralisation of powers at Holyrood (the Scottish Parliament), drawn simultaneously from both Westminster (UK government) and Scottish local authorities. Such a structure might even work, and would address one of the major post-devolution issues – “too much democracy”.

However, the most important lesson of the Edinburgh tram is not that such projects should be managed more centrally. Rather, that capital investments should be scaled to match the scale of the government trying to deliver them: The tram’s fundamental failing is that it is too expensive (and consequently risky) a mode for a local public transport project. If an inherently local policy can only be enacted by a non-local government, then (I argue) it is the wrong policy. Local policy ambitions need to be scaled down to what is reliably achievable at local level.

The issue will become critical for a future Scottish government with financial independence from the rest of the United Kingdom (a distinct possibility, given the current dominance of the Scottish National Party). Scotland represents approximately 10% of the UK population and economy. This implies that a truly independent Scottish government could only safely manage 1/10 the amount of risk that the UK government is capable of managing. Meaning, the most ambitious project an independent Scotland could achieve would be 90% less ambitious than a UK-backed project.

Scotland’s current transport projects are still reasonably modest, like reopening short sections of old railway line. But it is easy to imagine the Scottish Parliament’s ever-increasing ambition proposing grandiose projects that the UK might contemplate, but which could bankrupt Scotland – like a new high speed rail network: A core route serving Scotland’s largest 5 or 6 cities, with an under-sea tunnel to Lerwick in phase 2…

The challenge for Scotland is not to pretend to be big, to still be part of the British Empire, or a modern-day China. Quite the opposite: To deliver the same (or better) policy impact, while the maximum cost of individual projects is limited to a tenth what the population has learnt to expect. Overall expenditure would not be reduced – there would simply be far more, smaller, projects. Fortunately, there’s a lot more scope to improve the humble bus stop than you might think!

I suspect Scotland’s history already contains the best example of what can happen when a nation stops pretending to be big, and instead focuses on itself: The original Acts of Union (with England) removed colonial and military distractions, allowing Scotland to develop domestically as an intellectual and commercial 18th century power-house.

The management of chaos (of which financial risk is a part), may become one of the great challenges for structured 21st century government, regardless of size. Paradoxically, the easiest way to manage chaos is not to have to: To foster a system where no one part is too important to fail.

Appendix: Policy History of Edinburgh Trams

The idea for the Edinburgh tram was originally developed by the Scottish Office (the pre-devolution civil service responsible for administering Scotland), as part of a package of transport measures for south-east Scotland, to be funded by road pricing (tolls on car journeys within Edinburgh).

3 conditions made Edinburgh a perfect target for such ambitious transport policy initiatives:

1. Transport was (is) a constant source of annoyance for many city residents – a “political hot topic”.
2. Edinburgh’s geography – densely populated, but hilly – historically lead to above-average local public transport use, with bus travel remaining socially acceptable among Edinburgh’s “middle classes”.
3. Minimal democratic interference (the Thatcher/Major governments contained almost no Scottish Members of Parliament) allowed Scotland’s civil servants to focus.

The first Blairite UK government (1994) heralded both devolution in Scotland, and a shift in transport policy towards public transport. That combination should have been perfect for Edinburgh, except that the Scottish Executive (the devolved equivalent of the Scottish Office) became considerably more focused on managing opinions. Which in transport, has a tendency to result in nothing…

Instead, policy momentum was picked up by the recently liberated (from Lothian Regional Council during local government reorganisation) City of Edinburgh Council: Local ambition steadily grew, from “Greenways” (networks of bus lanes) in the early 1990s, through guided busways and radical parking policies, to strategies based around trams and road charging.

Charging proved too radical: Scottish government compelled Edinburgh to hold a referendum – but only on road pricing, not the trams that pricing had originally been intended to fund. Edinburgh’s population naturally rejected road pricing in isolation – Turkeys don’t vote for Christmas – and the tram project proceeded with traditional forms of funding.

In spite of (initially) primarily being funded by Scotland, the trams were local in scope (with no tangible benefit to anyone outside Edinburgh), and hence construction was “managed” by the council. This was done through Transport Initiatives Edinburgh, a council-owned company established to deliver Edinburgh’s light rail scheme – something which it has since struggled to achieve.

• Corporate deal, not the “founder buy-out” traditionally commonplace among gaming fansites. MMO-Champion was previously owned by Major League Gaming, already a multi-million dollar enterprise (by comparison, $46 million funding).
• Completes a duopoly (2 dominant businesses) in the core World of Warcraft “fansite” market – Curse and ZAM. While there are other large businesses and specialist niches on the fringe, none of those appear to be growing into the core WoW market.
• Exposes an intriguing driver of this market structure: Systems costs – the underlying technology and support costs. Intriguing because these were crucial in determining the market structure of far more traditional sectors of the economy, like groceries.

This article analyses the latest acquisitions and discusses the unseen importance of systems costs. On this page:

• The Other Online Game
• Curse vs ZAM
• Sold!
• Systems Costs
• Descent into Madness

The Other Online Game

New readers may be surprised to learn that there are millions of dollars involved in World of Warcraft-related websites. A Strange Game and Learn2Play, the new Real Money Trading provide an introduction.

My Map of World of Warcraft Online Communities showed the structure of this market in 2008. By 2010, all the communities in the core top-left quadrant of that map had become owned by large (typically multi-million dollar) businesses. The centre-ground is now commercially mature: In 2010 there is no big fansite that didn’t exist in 2008, while many of the large fansites in 2008 had barely existed in 2006.

All these large businesses are now diversifying into other games, because WoW hasn’t expanded beyond around 5-10 million players (excluding China), and inevitably this will fall at some point in the future (no video game can remain so dominant forever). Diversification is a tricky strategy because WoW continues (through strong marketing and new player retention) to attract customers, while competitors fail to challenge it. Indeed, its competitors are increasingly not even “Massively Multiplayer Online Games” – instead, browser games like Maple Story, or (Facebook) social games like Farmville.

Owners’ focus varies: For example, IncGamers has almost forgotten its roots as WorldofWar.Net, while WoWWiki is increasingly a platform to aggressively promote Wikia’s wider gaming portfolio. In contrast, Curse and ZAM still feel like they have World of Warcraft at their core, and continue to develop (or invest in) a range of services for World of Warcraft players.

So if World of Warcraft continues to be popular, and the core market remains mature, Curse and ZAM will tend to dominate popular fansite content. Other larger sites, without sufficient focus, will tend to get left behind (just like WorldofWar.Net), and smaller niche sites will tend to remain in small niches. This is why Curse and ZAM are interesting.

Curse vs ZAM

The Curse gaming network emerged from a World of Warcraft guild of the same name, originally as a database of addons – small scripts/programmes that can be run within the game. WoW addons remain its forté, with both developer and player-facing services. Curse started buying up WoW fansites in 2008, including MMO-Champion’s traditional competitor, World of Raids. World of Raids subsequently lost many of its visitors, while MMO-Champion thrived. Curse also attempted to launch a database website, in direct competition with ZAM’s Wowhead. Rather too direct: ZAM threatened legal action due to the similarity of WoWDB’s design.

ZAM was created from the acquisition of 3 large databases of gameplay information (Thottbot, Wowhead and Allakhazam) by some combination of Brock Pierce/Jon Yantis/IGE/Affinity Media (accounts vary). ZAM is understood to still to be privately owned. ZAM have also had some success with addons (especially WoW Interface), and have become interested in guides: Acquiring Tankspot’s video guides, writing event guides on Wowhead, even acting as a host for Deca/Alex Albrecht’s Project Lore.

In 2008, Curse’s ties with World of Raids would not have made it a natural ally of MMO-Champion. However, MMO-Champion hasn’t had flawless relations with ZAM either: Wowhead publicly appologised for “not appropriately crediting” the reuse of MMO-Champion’s content. The appearance of MMO-Champion’s own database probably didn’t help either.

While Major League Gaming does run World of Warcraft tournaments, MMO-Champion’s audience isn’t solely interested in tournaments. It is not surprising to see the site sold. And if my analysis is correct, there were only ever 2 potential buyers: Able to find the (undisclosed, but presumably substantial) fee, able to guarantee a reasonable amount of editorial freedom, and able to offer systems support:

Sold!

Independent fansite acquisitions typically involve:

• A creative founder suddenly discovers they can’t manage servers, staff, legal issues, or… and yet dealing with these things prevents them from continuing to do what they are good at. It might sound odd, by selling often isn’t about “the money”.
• An operating/managing/owning network that has experience of doing all the things the founder cannot, and can gain an “economy of scale” or market share or something useful from doing so: Typically by sharing expertise, management and information across many somewhat-similar websites.

At least that’s the ideal model. Some acquisitions are far more cynical or aggressive: Sites acquired solely for short-term advertising revenue, to resell a domain name, or to promote some dubious third party product or service.

This case is more complex, because the site hasn’t been independent for 3 years: It is difficult to judge all the reasons for a corporate acquisition (accountancy can often justify such decisions), but the main motivation for the sale appears to be technical. And not just visual issues, like the desire for fully-functional forums or a modern design. This is the most interesting reason Boubouille (the founder) gives:

“I think some people just underestimate the technical shitstorm behind each Beta patch or every single news. Having the backup of a WoW-focused company with tons of WoW-focused developers is a pretty huge thing for me.”

I hinted at the complexity of the process in the Adventures in DBC Files box. There are 2 problems:

1. Large amounts of regularly changing, but partial data, which must still be delivered accurately, as fast as possible. Hundreds of thousands of things, many inter-related.
2. Only a few “fansites” that need tools to analyse this data, and only a modest number of people with sufficient specialist knowledge to work with the data.

This is a lot like… groceries.

Systems Costs

Grocery stores’ (especially supermarkets’) competitive advantage lies in the rapid turnover of stock: The faster the store can buy merchandise from suppliers and sell it on to consumers, the sooner that cash will be available to buy more stock. Effective inventory (especially “supply chain”) management is key to profitability. That means lots of data, constantly changing.

But that’s not the only similarity: Each grocer gains advantage over competitors by developing “better” systems to manage inventory and related information. Sharing those systems with others would lose competitive advantage, yet ever-better systems cost more and more to develop. The only way to survive is for grocers to merge businesses, and share systems costs in a single merged business, even if they continue trading under familiar store names. It’s what forced US giant WalMart into Europe, and many European grocers to do the opposite. And for the whole industry to slowly agglomerate into a handful of huge global businesses.

Gaming fansites aren’t quite on the same scale as Wal-Mart. Yet. But the pattern emerging is similar: For example, programming expertise (and code) can’t be freely shared between competing fansites, because then everyone would produce precisely the same analysis at precisely the same speed, and any notion of competition would be lost. Yet several sites in the same network can share expertise much more freely between themselves – meaning fewer programmers or more flexibility.

While gaming sites are often profitable, most continue to make remarkably little money from so many customers: MMO-Champion may have 7 million monthly visitors, but it will struggle to earn a cent ($0.01) out of most of them. Most of that revenue disappears in operating costs, before a 40% profit margin (typically expected by early-stage venture capitalists) has been paid. On a fansite property that has probably already reached its maximum potential (barely 7 million people both play World of Warcraft and read English). So either this acquisition was part of an ill-conceived rush to spend venture capitalists’ cash (possibly before Games Workshop take it all in damages), or someone is planning to gain some serious efficiencies.

Descent into Madness

In theory almost anyone can setup a new WoW fansite. But in practice they won’t be able to compete anywhere except an undiscovered niche, because the technical, systems “barrier to entry” is now so high. The “fansite” market structure that emerges is the ultimate reflection on the complexity of the game itself: So much information, that individual websites can’t sustain themselves alone.

But then, rationally, none of this made any sense at the outset. As I previously concluded:

“Much of the third-party information is already coded into the game by the developers, obfuscated as game-play, discover by players, fed into third-party services, and then used to play the game by everyone else.”

We shouldn’t be concerned about systems costs because there should only be need of one system: The one in the game.

And I didn’t even mention that these third-party services are not licensed, but still tend to profit from the developers’ intellectual property. Or that such information might be better delivered in-game, not via websites. And that websites are preferred precisely because they can be operated commercially without permission. But then we’d have to acknowledge that these websites remain popular because they’re doing something game developers seem unable or unwilling to do.

While Curse and ZAM are still small businesses compared to mainstream game developers/publishers, both now have much to lose should anyone try to unpick this mess. Strange game indeed.