No market mechanisms accounted for in Plan S

The so-called cOAlition-S, consisting of 13 research funding organizations and three charitable foundations, recently launched it’s Plan S which has the aim of creating a major shift in publishing practices. (The “S” stands for “science, speed, shock, solution”). It was launched on 4th September this year, and the implementation plan was published 27th November. It has received a fair amount of critisism, and more importantly: a great deal of uncertainty and concern. (See our report here, and others here and here. For those reading Norwegian, some debate is collected here). While it is easy to agree on the ideal aims of the plan, it is a bit harder to judge the realism, unintended consequences and how the publishing industry will adapt. Here in Norway, about 1000 scientist have signed a letter demanding that the Norwegian research council would make a range of clarifications and do a thorough report on the plan’s consequences. So far, the Norwegian research council has refused to do so. Neither is there any report available on the homepage of cOAlition-S. In short: consequences have not been clarified, and certainly not in the open. I find this latter point quite ironic. 

While making research results openly available to the public and policy makers is obviously necessary, but as is generally recognized, not all means are necessarily justified by a good cause. Plan S is a specific plan demaning all funded research to be published in Gold Open Access. Here is the reasoning from cOAlition-S

Universality is a fundamental principle of science (the term “science” as used here includes the humanities): only results that can be discussed, challenged, and, where  appropriate, tested and reproduced by others qualify as scientific. Science, as an institution of organised criticism,  can therefore only function properly if research results are made openly available to the community so that they can be submitted to the test and scrutiny of other researchers. Furthermore, new research builds on established results from previous research. The chain, whereby new scientific discoveries are built on previously established results, can only work optimally if all research results are made openly available to the scientific community.

Publication paywalls are withholding a substantial amount of research results from a large fraction of the scientific community and from society as a whole. This constitutes an absolute anomaly, which hinders the scientific  enterprise in its very foundations and hampers its uptake by society. Monetising the access to new and existing research results is profoundly at odds with the ethos of science (Merton, 1973). There is no longer any justification for this state of affairs to prevail and the subscription-based model of scientific publishing, including its so-called  ‘hybrid’ variants, should therefore be terminated. In the 21st century, science publishers should provide a service to help researchers disseminate their results. They may be paid fair  value for the services they are providing, but no science should be locked behind paywalls!

I belive the sentence is at heart of the plan: “Monetising the access to new and existing research results is profoundly at odds with the ethos of science“. Easy to agree in principle. However, I do not see why monetising publishing is much better. Keep in mind that several big OA publishing houses are indeed commercial. Their source of income is the article processing charges (APC) instead of subscriptions. Thus, someone pays for publishing, one way or another. Plan S is portrayed as an ambitious plan to change the publishing models in science more broadly. There will be an end to “paying for research twice”, as they say. I am less convinced a new publishing model necessarily change the amount paid, though. But it will change who pays. Importantly, the plan does not do anything about the for-profit nature of publishing as such. cOAlition-S says it will make a report on article processing fees, but that is yet to come. They do say, though, that they belive prices will go down because of competition. Some market mechanisms are assumed to be involved, then.

Let’s assume Plan S works perfectly as intended: The whole publishing industry will be transformed and move to Gold Open Access. Subscription-based journals will perish. What is a likely scenario in this case?

Here are some basic conditions: 1) All costs and profits needs to be covered by article processing fees. 2) From the publishing houses’ perspective, the customers are no longer the libraries, but the individual researchers. 3) The research councils puts a ceiling on how much article publishing charge they will fund, 4) Existinc top-journals will switch to Gold OA or new top-journals will emerge, thus some kind of publishing hierachy will remain.

Then there is a basic question of how markets work: if demand goes up, so does the price, right? The very few generally recognized high-quality Gold OA journals will be able to charge more in article processing fees precisely because they are considered high quality. There is prestigue in publishing in the best outlets and such publications tend to have greater impact on the research community. That means both research institutions and individual researchers will be willing to pay for getting published there even if the research councils do not support it. Moreover, as long as such publications will help you land job (or tenure), researchers might even be willing to pay from their own pocket as it might pay in the longer run. For this reason, it is reasonable to expect that APCs for high-ranking journals might become very high while low-ranking journals might even be cheap. These are ordinary market mechanisms. However, Plan S states that compliance will be monitored and sanctioned, so that will restrict the prices, I suppose. Which leads us to the next point. 

The supply-side will probably also increase. Small profits on each single article might be compensated by publishing more. Actually, high profits on each article would also motivate for publishing more. Importantly, the founders of Plan S states that in the age of internet, publishing costs are low and there is no need to charge much. This also means that scaling up is cheap. All journals could just publish a lot more. Why reject papers unless they’re absolutely crap? Any mediocre paper might be published for some extra dollars. Why not? Indeed, some journals do so already.

I am concerned it will create an environment where unneeded journals will thrive: Journals that might not be quite predatory – although not necessarily far from it. The direct economic incentives will be to publish each paper, and too few papers will be rejected. There is already too much junk out there, the last thing we need is to lower the bar for getting published. I do not see how Plan S will handle any of such concerns. There is no credible plan for marked regulations. I do not mind some regulations, and I think that is generally important, but market regulations is not that easy! It is way too optimistic to think the plan will have the desired consequences when no specific concerns have been detailed. The cOalition-S homepage is currently strikingly void of information beyond the ten principles and the implementation plan, and both are pretty vague and in general terms. I am not sure it is a plan at all, but rather some high sounding language made into regulations. A plan that sets out to change the economic model for scientific publishing should pay close attention to the market mechanisms of which it interfers.

From what I have seen so far, cOAlition-S have not done any analysis of how the plan might work in that regard. At least, such analyses have not at all been anywhere close to Gold Open Access, if published at all. An open debate and explicit considerations should be made in writing and open for anyone to see. I do not at all understand why the cOAlition-S decides not to be open!

I would also like to point out that how compliance will be monitored and sanctioned is very opaque. What would it take to curb the market? It might be rather rough measures – and it might be directed at the individual researcher. It remains open how far the cOAlition-S will be willing to go to ensure compliance.

I belive there are lots of problems with our publishing system. I could go on a bit about that. However, I fail to see Plan S solving any of my concerns with the current system, and I am worried it will even increase some of the problems.

Risk-benefit considerations of male circumcision

Last week, the Norwegian socialist party (SV) added to their program that they wanted an age limit on male circumcision, at 15 years old. The socialist party is pretty small and the major parties do not agree, so there will be no change in Norwegian legislations on this. But there is one thing that have been bothering me – and does not really depend on one’s view on circumcision – which is the evidence on risk and benefits of circumcision.

A couple of years ago, male circumcision was much debated in Norway, and I considered writing something on this then. Never got around to it, though. I suppose similar discussions appear elsewhere from time to time, so this might be of some general interest. In the debates I’ve seen, those defending the practice frequently state that circumcision did no or little harm and actually have health benefits. The main scientific source referred to (at least in the debate in Norway) was the review of research published in Pediatrics in 2012. It was an official policy statement issued by the American Association of Pediatrics (AAP) and there was also a technical report which reviewed current evidence. As far as I am aware, this is still the AAP’s official stand on this matter. Importantly, the AAP do not generally recommend circumcision and it is not a considered a medical necessity. The main conclusion of the report is that the benefits outweight the risks. My concern here is what this means. The report discussed many potential risks and benefits of which some seems rather trivial to me, while others seems important. So how did the authors weight the risks and benefits against each other? I have no idea, because that is not described neither in the technical report or policy recommendation.

The report lists all evidence for and against a range of outcomes (most evidence is “fair”, and I think no evidence is “excellent”), but there are no judgments of how important each outcome is compared to the others. Clearly, it is hard to compare risks for very different diseases and complications, and I have no idea how this should be done. Both risks and benefits can be either trivial or important or anything in between. How do you weight a trivial harm with high probability against a serious benefit with extremely low probability? How do you weight a catastrophic complication of the procedure with extremely low probability against a certain but trivial benefit? How do you weight heterogeneous effects? These are very important considerations, and the panel have apparently made such considerations since they say these outcomes are weighted against each other. Still, the only argument is a long list of evidence on various outcomes. The considerations of trade-offs are highly informal, and I have no idea how they weight the unknowns.

I suggest that when writing a review and giving policy recommendations (on any topic) where you say something like “benefits outweight the risks“, at least you really need to get the following straight:

  1. If you say you have weighted benefits against risk, this better be clear and explicit. Just saying so is just not good enough. Not all risks and benefits are equally important and should be given different weights accordingly. Some risks should perhaps be avoided at all costs, some are quite manageable, and other risks are of no concern. Same applies to benefits: some are important, others are not. This needs to be made clear. Even if done badly would be better than nothing as it would allow important trade-offs to be discussed by others. If not explicit, I doubt it has been done at all.
  2. Moreover, some benefits can be achieved by other means without risk. With no such considerations, the weighting of pros and cons have not really been done.
  3. Risks which are possible but no estimates exist needs to be explicitly considered anyway. For example, if you do (as AAP does) state things like e.g. “there are no adequate studies of late complications”. You need to make it clear how lack of knowledge are given weight.  

My main point is just that the AAP lacks a clear argument for their policy recommendation. It seems pretty opaque.

The post Risk-benefit considerations of male circumcision appeared on The Grumpy Criminologist 2018-04-23 10:00:26 by Torbjørn.

Norwegian politics: Sorry seems to be the hardest word

There is political turmoil in Norway right now. The New York Times give a brief account in English in this link. It all started when our government suggested that citizens with dual citizenship who posed a threat to Norway could loose their citizenship, of which almost all political parties agreed. Seems reasonable. The debate was then whether this should be decided by a court or administratively by the ministry of Justice. The majority of the parlament (including the Labour party) said it should be done by a court.

That disagreement on the procedure for how to loose citizen rights made our former minister of Justice to post on Facebook that the Norwegian Labour Party is a treat to our nation, and that they give priority to terrorists rights over national interests, illustrating the point by including a photo of some scary looking jihadists. This might appear to be pretty silly and also clearly a false accusation given the actual political disagreement in this case was rather small. However, it is also the case that right-wing extremist conspiracy theories have been making similar claims for many years, and portraying the labour party as an enemy. This kind of conspiracy was an explicit motivation for Breiviks terror attack on Utøya, a camp for labour party youth organization, as well as the bomb outside the government building. Clearly, many were upset that the minister of Justice fuelled such thinking, willingly or unwillingly.

Everybody can do silly things. Most of us would apologize when we realize it, often straight away, and at least when we’re told. And certainly if the prime minister asks it from us. Our former minister of Justice did no such thing. It took six days until she was pretty much forced to apologize in the Parlament. As it was not at all convincing, she had to return to the podium four times to apologize, and still hardly anyone found it convincing. Maybe because she did not apologize the content of her statement. And maybe because she also refused to remove the post until it was discovered that the picture is owned by the Associated Press who did not permit it to be used in political campaigns.

Much can be said about this. But I will rather take a visual approach, making use of Microsoft API for analyzing facial expressions. What emotions do you express when you apologize if you do not really mean it?

Data

All debates in the parlament is filmed and available online here. I downloaded the film from parlament and used Windows Movie Maker to cut only the section where the Minister of Justice entered to podium the apologize the first time. I then used ffmpeg to split the film into snapshots, two each second, resulting in about 350 photos. I then used Azure Cognitive Services API for face detection and analysis, submitted each frame and collected the results to make some time series graphics.

Importantly: The Face API is a machine learning service that also can provide analysis of facial expressions. It scores eight emotions on a scale from 0 to 1, and the sum over emotions is 1. So each score can be interpreted as a probability. I could put some caveats here, but I won’t. Let’s just say it gives a pretty good idea of what emotions are expressed. The emotions are as follows: anger, contempt, disgust, fear, happiness, neutral, sadness and surprise. My expectation from an apology would be fairly high level of sadness. At least more sad than happy. Let’s see.

Results

The plot shows composition of emotions expressions by each picture frame (two pictures each second). The exact measures varies quite a lot, so this is the smoothed trend. There is perhaps less interesting development than I had hoped for, but on the other hand, I had no reason to expect a volatile emotional life on her part. Stable, but a slight shift from neutral to happiness (she was smiling a bit).

The facial expression is dominated by neutrality, at between 70 and 80%. Happiness accounted for the main bulk of the remaining, between 10 and 20%. Neutral and happiness accounted for about 90% throughout her speech. She scores pretty low on sadness, which might be one reason why nobody believed her apology.

The next plot shows the trends for the less dominant expressions during the same speech. I just think it is worth pointing out that her score on sadness went down towards the end while surprise increased. Maybe she sensed that this did not go down as well as intended.

I will get back to this later. Maybe it would be interesting to e.g. compare with others who have made apologies recently. How do you do this convincingly?

I could not make up my mind if this post should be a tribute to Elton John or Edit Piaf. So, Elton John got the title and Edith Piaf the coda.

 

The post Norwegian politics: Sorry seems to be the hardest word appeared on The Grumpy Criminologist 2018-03-21 15:08:20 by Torbjørn.
Social Media Auto Publish Powered By : XYZScripts.com