Archive for October, 2017

In 2015, climatologists at the National Oceanic and Atmospheric Administration (NOAA) published a study calling into question a global warming “hiatus”, in which the rate of global warming increase was supposedly slowing. Prior to this study, it was believed that the rate of increase in global warming from 1998 to 2012 was about 33% to 50% the rate of the previous decades. The NOAA study found that the slowdown was much less significant (and that there wasn’t potentially one at all). However, later that year, a whistleblower alleged that the paper ignored major issues in a “rush to publication,” including improper data storage, improper data selection, lack of transparency and not noting that the data was “experimental”.  This whistleblower was possibly Dr. John Bates, a former NOAA scientist who gave an interview in early 2017 with the Daily Mail where he reiterated many of the accusations (1. Rose- 2017). It’s worth noting that this Daily Mail article now has to open with a disclaimer that “The Independent Press Standards Organisation has upheld a complaint against this article” regarding journalistic integrity.

The allegations caused the NOAA to review the paper, and fueled many climate change skeptics to decry the organization and climate change in general. Lamar Smith, the head of the House Science Committee, used this as a talking point and attempted to subpoena the emails of the NOAA (in fairness, he was trying to subpoena them prior to the whistleblower stepping forward and frequently has tried to subpoena the emails of climatologists whose work he doesn’t agree). Initially, the charges made it seem that the NOAA study improperly used data, and perhaps even weighed it in order to get a particular outcome. However, the study actually based its data based on previously published papers, and several subsequent papers using independent data sources corroborated the NOAA’s findings (2. Wright 2017). In fact, the study methodology corrected bias in some previous data sets (previous data was aggregated from buoys and ships without accounting for measurement differences between the two. Ship data is less accurate than buoy data because ships can generate their own heat, and the new data puts more emphasis on the more accurate data source (3. Hausfather 2017). Granted, there has at least been one major paper in 2016 that still supported a potential “hiatus.”

In the end, Dr. Bates ended up clarifying that the data was not fraudulent, but he was not happy with some protocol breaches. However, it seems like his actual concern was not really about data at all, but rather that he believed the paper authors were trying to influence policy (specifically in regards to the Paris Climate Agreement). In his own words,

“You really have to provide the most objective view and let the             policymakers decide from their role. I’m getting much more wary of scientists growing into too much advocacy. I think there is certainly a role there, and yet people have to really examine themselves for their own bias and be careful about that.” (4. Waldman- 2017).

While we should always be examining our own bias, the scientific community has a right to influence policy; it’s illogical to disseminate important information, and then leave all the policy decisions to people who deny your findings with no factual basis. If the data was improperly obtained or if the article had incorrect results due to rushing, this would be a different story. However, neither of these are the case. I was recently reminded of this scandal when someone was posting this as proof not to trust climatologists, so the allegations of impropriety have real world effects in shaping how voters view climate change. Additionally, much of the groundwork for Paris had been laid out prior to the 2015 paper being published, so it’s hard to say whether it even had any effect.

The reality is in the end, working quickly to get out a paper that confirms what the scientific community already agreed upon was not much of a political act. As scientists, we not only have to understand what we are researching, but also how it will be used; this also extends to critiques we make. The irony is that in attempting to stop other scientists from advocating for policy, Dr. Bates drew far more negative political attention to the subject at hand. He was the one who politicized the paper.

 

Works Cited

1. Rose, D (2017, Feb 4). Exposed: How world leaders were duped into investing billions over manipulated global warming data. Daily Mail.

http://www.dailymail.co.uk/sciencetech/article-4192182/World-leaders-duped-manipulated-global-warming-data.html

2. Wright, P (2017, Feb 9). The Data is Right: Climate Change is Still Real. Weather.com

https://weather.com/science/environment/news/climate-change-noaa-controversy-study

3. Hausfather, Z (2017, Feb 5). Factcheck: Mail on Sunday’s ‘astonishing evidence’ about global temperature rise. CarbonBrief

Factcheck: Mail on Sunday’s ‘astonishing evidence’ about global temperature rise

4. Waldman, S (2017, Feb 7). ‘Whistleblower’ says protocol was breached but no data fraud. E&E News

https://www.eenews.net/stories/1060049630

Please join us for the NLP Seminar on Monday, October 30, at 4:00pm in 202 South Hall.  All are welcome!

Speaker: Christopher Potts (Stanford Linguistics)

Title:  Enriching distributional linguistic representations with structured resources

Abstract:

One of the most powerful ideas in natural language processing is that we can represent words and phrases using dense vectors learned from co-occurrence patterns in text. Such representations have proven themselves in many settings, and one might even argue that they make good on a common intuition among linguists: that words tend to be incredibly complex and related to each other in all sorts of subtle ways. However, co-occurrence patterns alone tend to yield only a blurry picture of the rich relationships that exist between concepts, which raises the question of how best to incorporate additional information from more structured resources. This talk will explore methods for achieving this synthesis, with special emphasis on the retrofitting method pioneered by Faruqui et al. (2015), in which existing representations are updated based on their position in a knowledge graph. I’ll describe and motivate a generalization of Faruqui et al.’s framework that explicitly models graph relations as functions (Lengerich et al. 2017), and I’ll discuss some potential pitfalls of retrofitting (Cases et al. 2017). My overall goal is to stimulate discussion about how to obtain semantically nuanced distributed representations that are useful in diverse tasks.

( Slides )

References:

Cases, Ignacio; Minh-Thang Luong; and Christopher Potts. 2017. On the effective use of pretraining for natural language inference. Ms., Stanford University. https://arxiv.org/abs/1710.02076

Faruqui, Manaal; Jesse Dodge; Sujay K. Jauhar; Chris Dyer; Eduard Hovy; and Noah A. Smith. 2015. Retrofitting word vectors to semantic lexicons. NAACL. http://www.aclweb.org/anthology/N15-1184

Lengerich, Benjamin J.; Andrew L. Maas; and Christopher Potts. 2017. Retrofitting distributional embeddings to knowledge graphs with functional relations. Ms., Carnegie Mellon University, Stanford University, and Roam Analytics. https://arxiv.org/abs/1708.00112

Fall 2017 Test

October 12th, 2017

Hi there, everyone! This is a “test” post to ensure that the process is working as intended and that everyone should have access to create posts of your own!