skip to main content


Title: TESTING TRANSPARENCY
In modern democracies, governmental transparency is thought to have great value. When it comes to addressing administrative corruption and mismanagement, many would agree with Justice Brandeis’s observation that sunlight is the best disinfectant. Beyond this, many credit transparency with enabling meaningful citizen participation. But even though transparency appears highly correlated with successful governance in developed democracies, assumptions about administrative transparency have remained empirically untested. Testing effects of transparency would prove particularly helpful in developing democracies where transparency norms have not taken hold or only have done so slowly. In these contexts, does administrative transparency really create the sorts of benefits attributed to it? Transparency might grease the gears of developed democracies, but what good is grease when many of the gears seem to be broken or missing entirely? This Article presents empirical results from a first-of-its-kind field study that tested two major promises of administrative transparency in a developing democracy: that transparency increases public participation in government affairs and that it increases government accountability. To test these hypotheses, we used two randomized controlled trials. Surprisingly, we found transparency had no significant effect in almost any of our quantitative measurements, although our qualitative results suggested that when transparency interventions exposed corruption, some limited oversight could result. Our findings are particularly significant for developing democracies and show, at least in this context, that Justice Brandeis may have oversold the cleansing effects of transparency. A few rays of transparency shining light on government action do not disinfect the system and cure government corruption and mismanagement. Once corruption and mismanagement are identified, it takes effective government institutions and action from civil society to successfully act as a disinfectant.  more » « less
Award ID(s):
1655459
NSF-PAR ID:
10187416
Author(s) / Creator(s):
; ;
Date Published:
Journal Name:
Northwestern University law review
Volume:
114
Issue:
5
ISSN:
0029-3571
Page Range / eLocation ID:
1263-1334
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)
    The Neuronix high-performance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from low-end consumer grade devices such as the Nvidia GTX 970 to higher-end devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely compute-intensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3] as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floating-point calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should produce identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly well-known issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely time-consuming for algorithm research in which a single run often taxes a computing environment to its limits. Well-known techniques such as cross-validation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when re-running the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Cross-validation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Large-scale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPU-accelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are non-deterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multi-threading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32-bit since Python defaults to 64-bit. We try to avoid using 64-bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [11-13]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues. 
    more » « less
  2. Douglas Kriner (Ed.)
    Abstract

    Transparency is essential for public awareness of government activity and in holding officials accountable for their behavior and decisions. However, executives understandably have a desire to maintain autonomy over the flow of information outside of the executive branch, which can sometimes lead to a clash between government openness and executive control over information. This article investigates executive branch transparency along two specific dimensions: threats to the preservation of government information, and the judicial monitoring of executive branch transparency. Federal law directs executive branch actors to preserve government information. Statutes and regulations detail governing policies as well as guidelines that apply to every presidential administration. However, we consider each rule and recommendation through the lens of the Trump administration to gain greater clarity and precision. The Trump administration proves a useful focus for our empirical analysis because of its recency as well as multiple reports of its improper handling and preservation of records. We lay out and develop a typology of threats to record preservation. Preservation conflicts have emerged across multiple presidential administrations and have become increasingly common and visible. Judicial monitoring of executive branch transparency has become a crucial tool to address the withholding of records. We show that federal judges can play a vital role in defining the boundaries of government transparency through their decisions in Freedom of Information Act (FOIA) litigation. The U.S. Department of Justice oversees and often leads the government's defense against FOIA actions. We find that federal judges are most likely to require executive branch responsiveness and disclosure when the judges' underlying policy preferences are most distant from that of the administration. Our findings should inform reforms to protect government transparency.

     
    more » « less
  3. Abstract

    Collective action among public officials is necessary for the effective delivery of many social services, but relatively little is known about how it can be fostered through policy reforms. In this article, we compare cooperation among public officials within decentralized versus centrally‐administered municipalities in Honduras. Leveraging a quasi‐experiment in health sector reform, coupled with behavioral games and social network surveys, we find that decentralization is associated with greater cooperation. When they are able to communicate, health sector workers in decentralized municipalities contribute more to a public good than their centrally‐administered counterparts. This increase in cooperative behavior results in part from the decentralization reform engendering greater numbers of interactions and stronger ties across different levels of government. These findings indicate that institutional reforms like decentralization can favorably reconfigure patterns of social interactions across public organizations, which is an important component of administrative capacity in developing countries.

     
    more » « less
  4. With support from the National Science Foundation’s Division of Undergraduate Education, this five-year project led by a two-year HSI seeks to provide underrepresented students with mentored work experiences in computer information systems. Students will have access to on-campus work experiences and internships in businesses and industries. It is anticipated that some examples of potential student projects include mobile application development, cybersecurity, and computer support. It is expected that these experiences will increase undergraduate student interest, persistence, and success in computer information systems, as well as in STEM more broadly. To ensure that they are well-prepared for and gain the most from their work experiences, students will receive training on employability skills such as communication, teamwork, and project management. In addition, during their work experiences, students will be mentored by faculty, industry professionals, and peers. To strengthen the capacity of faculty to serve all students, including Hispanic students, the project will provide faculty with professional development focused on equity mindset. This framework to provide mentored work experiences will be developed and piloted at Phoenix College, in the computer information technology department and eventually expanded to other STEM fields at the institution. Following this, the project also intends to expand this framework four other two-year HSIs in the region. Through this work, the project aims to develop a replicable model for how two-year institutions can develop work experiences that foster increased student graduation and entry into STEM career pathways. This project, which is currently in its first year, seeks to examine how a curriculum that integrates cross-sector partnerships to provide work experiences can enhance STEM learning and retention. Using mixed methods and grounded theory, the project will expand knowledge about: (1) the impact of cross-sector partnerships that support work-focused experiential teaching and learning; (2) systematic ways to maintain and better use cross-sector partnerships; and (3) the degree to which a model of work-focused learning experiences can be adopted at other two-year HSIs and by other STEM fields. Baseline data about Hispanic serving identity at the pilot institution has been collected and assessed at the institutional, departmental, and for different educator roles including faculty, support staff, and administrative leaders to produce inputs towards developing a detailed plan of action. Early results from baseline data, visualizations, and planning responses will be reported in the submission. Expected long term results of the project include: development of sustainable mechanisms to foster cross-sector partnerships; increased student retention and workforce readiness; and measurable successes for STEM students, particularly Hispanic students, at two-year HSIs. 
    more » « less
  5. A Novel Community Engaged System Thinking Approach to Controlling Nutrient Pollution in the Belize Cayes Nutrient pollution (anthropogenic discharge of nitrogen and phosphate) is a major concern in many parts of the world. Excess nutrient discharge into nutrient limited waters can cause toxic algal blooms that lead to hypoxic zones, fish die-offs, and overgrowth on reefs. This can lead to coral reefs being more vulnerable to global warming and ocean acidification. For coastal communities that depend of fishing and tourism for their livelihood, and for reefs to protect coastlines, these effects can be devastating. A major source of nutrient input into the aquatic environment is poorly treated wastewater from Onsite Wastewater Treatment Systems (OWTS). When properly sited, built, and maintained conventional OWTS are great for removing fats, grease, biological oxygen demand (BOD), and total suspended solids (TSS), but they are rarely designed for nutrient removal and commonly have high nutrient levels in their effluent. This study investigates the factors that influence the performance of OWTS, the Caribbean region’s most common type of treatment technology, in the Belizean Cayes where salt water flushing is common. Using mass-balance-based models for existing and proposed OWTS to predict the system’s performance under various conditions, along with OWTS’ owner, maintainer, and user input, a novel community engaged system thinking approach to controlling nutrient pollution will be developed. Key model performance metrics are concentrations of nitrogen species, BOD, and TSS in the effluent. To demonstrate the model’s utility, a sensitivity analysis was performed for case studies in Belize, estimating the impact on nutrient removal efficiency when changes are made to variables such as number of daily users, idle periods, tank number and volume, oxygen concentration and recirculation. For the systems considered here, strategies such as aeration, increased biodigester tank size, addition of aerobic and anoxic biodigesters, recirculation, addition of a carbon source, ion exchange media is predicted to decrease the effluent nitrogen concentration, and integration of vegetation for nutrient uptake both on land and in the nearshore environment. In a previous case, the addition of an aerator was predicted to decrease the effluent ammonium concentration by 13%, whereas increasing the biodigester tank size would only decrease the effluent ammonium concentration by ~7%. Model results are shared with system manufacturers and operators to prioritize possible modifications, thereby optimizing the use of finite resources, namely time and money, for costly trial-and-error improvement efforts. 
    more » « less