Navigating a datascape: challenges in automating environmental data disclosure in China

Abstract

This article examines the current trend toward solving issues of procurement and processing of publicly disclosed pollution source data in China, where this data is characterized by heterogeneity and lack of standardization. Through ethnography and software analysis, the article examines the hidden labor entailed in automation using the case study of a Chinese e-NGO. We identify the bulk of this labor in “datascape navigation”, or the practices needed to locate, acquire and process the desired information within the infrastructure enabling the circulation of the data. The aspects of this labor are examined in relation to two data flows: enterprise environmental supervision records and information about real-time emissions. We identify several forms of unpredicted human and non-human labor entailed by both unsuccessful and successful automation attempts. We conclude that the labor involved by software automation of environmental data procurement and processing can critically impact environmental disclosure timing and quality.

Keywords:

• China
• environment
• data
• software
• automation

Acknowledgements

The author thanks the IPE’s staff and its director Ma Jun for their strong and continuous support and the two anonymous reviewers for their precious feedback, Christine Leuenberger of Cornell University, Valérie November at LATTS/Ecole des Ponts ParisTech, and Li Lulu of Renmin University for their guidance. This article includes text co-authored with Basile Zimmermann of the University of Geneva, as part of a monograph manuscript on the project “Website Analysis in China: The Pollution Map of the Institute of Public and Environmental Affairs” (CHIPOMAP) upon which this research is based. It is used here with his agreement.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed here.

Notes

1 The exact terminology has not yet reached a consensus in the scientific community: for instance, some commentators include ‘data procurement’ in data wrangling (Terrizzano et al. 2015).

2 Consider, for example, emission data going from a CEMS sensor to a processing unit via cable, from there to an institutional database via UMTS wireless connections, and henceforth to a disclosure website via database connections and TCP/IP packet switching protocols, and from there to the screen where actor A is seeing it, via an Ethernet port, a ×86-family processor, a video card and a VGA cable. But also, consider actor B driving to a library, going to the desk, asking the librarian about the location of a report, walking through the library, pulling the volume, perusing it and copying the information on a notebook. Both actors are traversing datascapes with different sociomaterial features. Needless to say, data flows are never this linear and tend towards more rhizomatic topographies, forking into multiple streams at different junctures.

3 The IPE allowed us to observe and partake of its activities on a daily basis, and generously provided access to employees, source code, procedures and prototypes; it did, however, retain direct access to its database, beyond the interfaces available through the organization’s website http://www.ipe.org.cn.

4 Other kinds of entities are included in EPBs’ lists, such as restaurants. However, they are not included in the IPE database.

5 This has been a key problem for the IPE’s “pollution map” from its inception. To chart companies, the IPE needs their latitude and longitude. Such coordinates are seldom included in violation records (or any other public record IPE could access). When company addresses are on record, they are often imprecise (also due to the uncertain street naming in China’s industrial spaces), thus the use of reverse geocoding functions of services such as Baidu Maps or Google Maps to obtain the coordinates. This forced the IPE to establish a costly process in which volunteers would physically go in front of the factory and take geocoded pictures of the gates. Then, further work was needed to correct the GPS coordinates due to China’s encrypted positioning system (Tarantino and Zimmermann, 2017). As a result, only a fraction of factories would actually appear on the IPE’s map. This has been a concern of the Ministry itself, which as of 2018 is asking factories to provide, among other things, exact geographic location in coordinates.

6 For example, in 2017, 620,000 records have been disclosed by 31 provinces, with information proceeding from rounds of inspections on illegal construction projects (建设项目清理). In 2015, some provinces—such as Shandong—also started publishing environmental credit records in lists, releasing data about thousands of factories at the same time.

7 At the time of our observation, particularly relevant were the Work Notice on Strengthening the Disclosure of Pollution Monitoring Data, and the (Pilot) Measures on the Disclosure of Self-Monitoring Data of Companies under Special Supervision (国家重点监控企业自行监测及信息公开办法（试行）) from July 30, 2013. Also the 2015 Air Pollution Prevention and Control Law 中华人民共和国大气污染防治法(2015修订) contains requirements in this regard.

8 Such companies are included in a list published annually by the Ministry since 2004 (年国家重点监控企业名单).

9 Standard for Data Communication of Pollution Emission Auto Monitoring System (中华人民共和国环境保护行业标准).

10 The 2013 release of the standard prescribes “special emission limits” (特别排放限值) for enterprises located in areas characterized by high development density or a fragility of the local environment: “In places where the density of the exploitation is already high and where the capacity of the environment to bear its weight has been weakened; or places where the capacity of the environment is lower, the ecosystem is fragile, and serious environmental pollution issues can occur easily, it is [then] necessary to adopt [the standard of] area under special protection measures.”

11 See http://www.mep.gov.cn/gkml/hbb/bgt/201603/t20160311_332712.htm

Additional information

Funding

This work was supported by the Swiss National Science Foundation under Grant number 153291. Supplementary funds were granted by the Confucius Institute of the University of Geneva.