So what would an oil contracts database look like?
Imagine a world in which extractive industry contracts were routinely published, from Article 1 definitions to Annexes with long lists of GPS data points of contract areas and everything from production sharing splits to management structures inbetween. Imagine the typical corporate arguments of breach of confidentiality agreements and conflict with commercial interest had been trumped, as they should be, by the fact that one of the two partners to the agreement is a sovereign state, and that a consensus now accepts that the public interest in publication, for billions of consumers as well as the citizens of the world’s producing countries, both outweighs and is compatible with reasonable commercial interests.
How would we make sense of the world’s oil and mining contracts? How could we get from a mass of jumbled heterogenous data to building a resource which converted it into meaningful knowledge and became a source of reference for those seeking fair negotiations leading to stable contracts?
Let’s work backwards from the end goal. As of now, the first issue is that, as has been noted elsewhere on this site, the data set is limited and probably unrepresentative: only five or six jurisdictions have currently published contracts and most of them are early stage or marginal producers. However much larger numbers of contracts are “semi-published” – available in expensive commercial databases such as Wood Mackensie. The RWI report which looked at contract transparency was able to gather information about over 150 contracts, something which in itself might lead us to consider that the claim that publishing contracts is dangerously revolutionary is in fact, merely disingenuous. It’s OK to publish for and among elites, apparently. We don’t see anybody chasing down these business publishers or threatening to take them to court.
But the value in a database of contracts designed for public use will be as much, maybe even more, in the metadata as in the original data of the contracts. It might be possible, without breach of copyright, to refer to contracts in the commercial databases and provide at least basic information on them. Even just the list of contracts, their formal titles, and the principals involved in negotiating them would be distinct progress. A major oil project could have as many as 60 separate legal agreements and we could at least convert the situation we have now of total unfamiliarity into a series of “known unknowns”, as Donald Rumsfeld would have it.
Plus, there are many more model contracts available than signed ones. At OpenOil we are advocates of full contract transparency but there’s no doubt you can learn things from model contracts. For example, if you were interested in the potential environmental consequences of the development of Iraq’s oil industry, you could read in the model contracts for the first two rounds of service agreements what procedures are stipulated to handle spills, and that, for example, “international best practice” is referred to several times without ever being unambiguously defined, creating lots of wiggle room in terms of liability in the event of any future spill. Similarly, if you were to read the model EPSA contracts in Libya, you would discover that a confidentiality agreement applies only to the company side – meaning that the new government would have no case to answer,at the International Chamber of Commerce in Paris or anywhere else, if it chose to publish the full contracts, whatever corporate executives glibly assume.
There are three other issues in how to build a contracts database. One is technical: if the value is as much in the metadata as the data, how can a system be structured which allows the accretion over time of quality commentaries and annotations, what we might call “Crowdsourcing Lite” from groups of professionals who want to give back? This is complex but has been largely solved in other domains, there’s no need to reinvent the wheel.
If we stay within the legal domain, LexisNexis’s free service LexisOne offers one model. The Pennsylvania Supreme Court, for example, has just ruled against an appeal by Hoffman Mining for an exemption to a local-level ruling which forbids them from surface mining within 1,000 feet of residences. The company wanted to cut the 1,000 feet exlusion zone to 300 feet, claiming there is $8 million worth of coal in the 700 feet difference. LexisOne carries a brief summary which then links to a second level, a legal alert by law firm Steptoe & Johnson, and finally, the deepest level, the full text of the ruling issued by the court.
LexisNexis of course is a company with an incumbent business model to protect, so it might or might not be possible to negotiate good terms to adapt such a system. But the problem of opening texts to peer review and multiple layers of annotation has also been solved in many academic contexts.
My own favourite is the Perseus Project, which has digitised all the texts of Latin and Greek antiquity, together with huge volumes of translation, commentaries and concordances which can all be turned on or off at user discretion. Go to the opening of Plato’s Republic, for example, and alongside the Greek text you can turn on English translation line by line, James Adam’s comprehensive 1902 commentary, and every word of the original text is a link to a page which is Perseus’s own concordance for where else it appears in the entire corpus of texts it is storing. Tufts University, meanwhile, which hosts the Perseus Project, is running courses in Text in the Digital Age and the Text Encoding Initiative, which has been running for nearly thirty years, has evolved an open standard of XML which is widely used to annotate texts. So there are open source options which, while they would not be trivial to adapt and set-up, clearly point to a long-term model for a contracts database which need not be constrained by any proprietary system.
The second key aspect of making such a database usable where it counts is the question of translation. Here also technical innovation could bring surprising results by focusing its own limited core resources first on the metadata rather than the core texts, and then using a tiering system to prioritise translation requests, with professional translation available at the top for urgent or much demanded segments, leading down through various stages of machine-assisted translation for lower priority. It’s worth remembering that commercial software translation packages are an order of dimension better than the machine translation most people are now used to dealing with, Google Translate, as long as you are dealing with controlled vocabularies and contexts, which would be the case here.
The third challenge is the human and political one of encouraging the right kind of input from the right kind of people, lawyers, accountants, geologists, engineers and managers. This is harder and longer-term. But there is a broad range of “respectable” institutions interested in pursuing such an idea, each of them with outreach to their own constituencies. Once a system was in place that offered the low-friction transactionality we are used to on digital platforms, momentum to annotate, comment and explain these contracts could gather quite quickly as long as it was handled in the right way. But in 2012, knowledge of how to nurture online communities is now two decades old. It would simply be a matter of adaptation.
There’s no reason, actually, why even a free database might not be a commercially sustainable operation. Running costs would be relatively low. And think back to Steptoe & Johnson contributing their alerts and the text of the Pennsylvanie ruling to LexisNexis’s free service. If a database got traction with governments and others looking at management of industries worth hundreds of billions of dollars, you could easily see a situation in which law firms and others would buy paid opinion slots, like Google Ads, or premium services such as customised print publishing were offered on top of the free website.
Each of the links of the chain are doable – they just need stringing together.