Offshore A-Z is a digital repository of companies operating across major tax havens. Offshore A-Z exploits the inscrutability of such companies to speculate on their possible nature and intent. Each company listed on here is real, but the information surrounding these consist of both actual records, obtained by dmstfctn, and speculative images and mission statements, generated by algorithms.

Tax havens, or secrecy jurisdictions, employ laws and corporate regulations different from those of most jurisdictions, using secrecy as a prime tool. Furthermore, the agents operating offshore on behalf of clients elsewhere have been observed by dmstfctn to willfully introduce mistakes in the records held on the companies they help forming. (Offshore Investigation Vehicle)

When offshore companies are but a bundle of papers held in private storages, these practices contribute to a number of possible configurations of any given offshore company, whose records are at once factually correct and conveniently flawed.

Offshore A-Z plays with the idea that truth is not one nor static, and that it does not need to be upheld equally at all times or in all fora.

Obtaining the data

The data contained within Offshore A-Z comes from a number of sources. Prior to the Offshore Leaks, the snippets of information available on offshore companies were largely confined to the outdated online registrars kept by tax havens, as an empty gesture towards transparency. Employing data scraping techniques, dmstfctn extracted records on hundreds of thousands of companies appearing in a number of such registrars.

Yielding varying amounts of information, and naturally full of gaps - sometimes full company details and descriptions, sometimes little more than a name - the records were cross referenced with data from the ICIJ’s Offshore Leaks as well as Financial Secrecy Index scores from the Tax Justice Network, and compiled into a single searchable database.

Training the algorithms

Algorithms were then trained to fill in the gaps in information. Using keywords from the company’s records as a trigger, the algorithms associate images with each nondescript company and complete missing fields by guessing possible incorporation dates or by generating corporate mission statement via a simple neural network. The coherency and accuracy of the neural network is directly undermined by the opaque source material, or incomplete input records. Finally, companies previously named in the Offshore Leaks ( for example) are hyperlinked to the relevant entry on the ICIJ database, allowing user to investigate further.

Compiling the template

The resulting information is autonomously built into a Web 2.0 page, employing a visual grammar similar to that used by existing registrars of companies. However, under intense scrutiny, the template mealts away revealing warm hues and floating forms.

This cycle of actions is repeated for each unreleased company found in the database and publicly announced by a Twitter bot.

Zooming in

Among the companies listed on the Offshore A-Z, a few appearing to be embedded in complex offshore vehicles were further investigated, resulting in three stories focusing on the funding behind Leave.EU, the corporate tax-avoidance of Barclays and the colonial origins of HSBC.

Offshore A-Z is a project by dmstfctn.