the money and control needed to coordinate all the players in the innovation system and hold them accountable for working toward a common goal.

was due to the patient-activists.

societal influences but instead by being brought, carefully and appropriately, into a direct, open, and intimate relationship with those influences.

But much of the problem can be traced back to a bald-faced but beautiful lie upon which rests the political and cultural power of science. This lie received its most compelling articulation just as America was about to embark on an extended period of extraordinary scientific, technological, and economic growth. It goes like this: Scientific progress on a broad front results from the free play of free intellects, working on subjects of their own choice, in the manner dictated by their curiosity for exploration of the unknown.

nothing made by humans was orbiting the earth

Outside efforts to guide the course of science would only interfere with its free and unpredictable advance.

Publicly available preprints provide an opportunity for authors to obtain feedback beyond the few scientists who see the manuscript during peer review.

inally, we wish to draw a distinction between data that is machine-actionable as a result of specific investment in software supporting that data-type, for example, bespoke parsers that understand life science wwPDB files or space science Space Physics Archive Search and Extract (SPASE) files, and data that is machine-actionable exclusively through the utilization of general-purpose, open technologies.

The discovery of this new lesion might call some of their conclusions about the functions of the medial temporal lobes into question and require a re-­examination of all that old data.

Yeah, but it’s not peer-­reviewed, for one thing. That’s important. The stuff that’s published is good stuff. Peer-­reviewed. You can believe it.

Nobody’s gonna look at them.Me: Really? I can’t imagine shredding the files of the most important research subject in history. Why would you do that?Corkin: Well, you can’t just take one test on one day and draw conclusions about it. That’s a very dangerous thing to do.

I had known Corkin since I was a kid. Remember her friend, the surgeon’s daughter, the one she tin-can-­telephoned with as a little girl? That’s my mom. The surgeon who performed the experimental operation on Henry was my grandfather. When I was growing up, Corkin was a staple at my mom’s dinner parties. We had met many times before.

because she became such a dominant authority figure in his life.

Even as a nonscientist, I couldn’t help noticing that some of the unpublished data I came across while reporting my book went against the grain of the established narrative of Patient H.M. For example, unpublished parts of a three-page psychological assessment of Henry provided evidence that even before the operation that transformed Henry Molaison into the amnesiac Patient H.M., his memory was already severely impaired.

I don’t think scientists would say that.

But you know, even what’s published — as you know, if you look at the papers, in some sense each paper is just the tip of the iceberg of the work that was done, and the work that was done — all that data floating around underneath — it seems to me that so much of that would be valuable to preserve. That people really may want to go back and review —

This would make it easier for scientists to continue their analysis, mining Henry’s brain for any last revelations it contained.

I had tracked down and spoken with Henry’s closest living relatives, and some were surprised and disturbed to learn about the things Corkin and her colleagues did with their cousin while he was alive and about the fight over his brain that took place after his death.

I asked Corkin whether she was aware that when Mooney became Henry’s conservator, one of Henry’s first cousins, Frank Molaison, was living nearby — his actual next of kin — and had not been consulted. I mentioned that his name should have made him particularly easy to find.

allow U.C.-­Davis certain rights to use and distribute the tissue owned by M.I.T. and M.G.H.” As far as Corkin was concerned, she and her colleagues owned Henry’s brain, period, and Annese had no say in the matter whatsoever.

The paperwork — the document she passed around the table at the meeting in New York — was only one page, and the crucial part took up just two sentences: “I, Thomas F. Mooney, am the court-­appointed guardian of the person of Henry G. Molaison. I also presently am Henry G. Molaison’s closest living next of kin, and as such I am entitled by law to control Henry G. Molaison’s remains upon his death.” The lines were followed by a signature and a date: Dec. 19, 1992. Corkin had arranged for Mooney to apply to become Henry’s conservator earlier that year. A probate-court judge, taking Mooney to be Henry’s closest relative, approved the conservatorship. One of Mooney’s first acts as conservator was to donate Henry’s brain to Corkin and her colleagues. He also consented, with Henry’s assent, to the continuation of the experiments Corkin wished to conduct on Henry while he remained alive. The problem was, Mooney wasn’t actually Henry’s next of kin.

His ego had given him a sense of entitlement.

The frontal lesion would stay in the paper, but it wouldn’t be featured as prominently as it was in earlier drafts.

(we ascertained it was present even in the 1992–93 M.R.I. scans)

As one of the paper’s anonymous peer reviewers pointed out, “much of the neuropsychological literature on H.M. has made the case that so-­called frontal function was intact.”

Because it’s a decision by the people who own the tissue.

Annese was wary of handing over data before they had an agreement for how his work would be credited.

Back in San Diego, Annese kicked off what he dubbed Project H.M. with what was possibly the most successful publicity stunt in the history of neuroscience.

but my reporting also eventually raised serious questions about Henry’s treatment after he left the operating room, during the decades he spent as a human research subject, as well as in the eight years that have passed since his death.

but I abandoned that effort after Corkin presented me with a confidentiality agreement stating that M.I.T. would allow me access to the “research project entitled ‘The Amnesic Patient H.M.’ ” only if the university had editorial control over anything I intended to publish.

Later, reflecting on that moment, Corkin could think of only one word to describe her feelings. She was, she wrote, “ecstatic.”

Now she was having one last encounter that only she would remember. The men carefully pulled out Henry’s brain, and Corkin gazed at it through the glass, marveling at this object she had spent her career considering at one step removed.

That paper documented Henry’s gradual improvement over a three-day period on a difficult hand-eye coordination task.

In that pantheon of illuminatingly broken men and women, Henry stands apart. It is difficult to exaggerate the impact he has had on our understanding of ourselves.

Corkin built much of her subsequent career on the back of her privileged access to Henry and became both his gatekeeper — fielding requests from other scientists who wished to meet him — and his chief inquisitor.

She grew up across the street from the neurosurgeon who operated on him and was close friends with the surgeon’s daughter.

FAQ for data repositories

(?)

Some of the libraries contributing to the foundation of Slicer were designed in close collaboration and often share the same developer community. These libraries, including CMake, ITK, VTK and CTK, are distributed as part of the National Alliance for Medical Image Computing (NA-MIC) Kit [153], which are actively supported by the NA-MIC research community28. Many popular packages, e.g., ANTs, MindBoggle, ITK-SNAP, DTIPrep, and MITK are also based on the NA-MIC Kit. NIPY29 and NeuroDebian30 are another two major research communities for neuroimaging research and platform development. To promote open science, neuroimaging tools and resources are always shared to other community members, usually through the INCF31 and NITRC32 forums.

This data reduction—taking a rich, multivariate dataset and summarizing it for publication using measures of central tendency, confidence intervals, p-values, and effect sizes—removes the opportunity for future scientists to apply new algorithms, methods, and transdisciplinary ideas that could yield unforeseen insights and discoveries

I outline the overlooked benefits of data sharing: novel remixing and combining as well as bias minimization and meta-analysis.

Such developments are addressing concerns regarding credit and help motivate data curation and contextualization.

Rather than arguing for a centralized, large-scale data repository, I am advocating for a more organic development wherein we, institutionally, encourage the growth of a data ecosystem.

Where datasets are hosted in public repositories that provide datasets with Digital Object Identifiers (DOIs), we encourage these datasets to be formally cited in reference lists. Citations of datasets, when they appear in the reference list, should include the minimum information recommended by DataCite and follow journal style. For example:Hao, Z., AghaKouchak, A., Nakhjiri, N., Farahmand, A. Global Integrated Drought Monitoring and Prediction System (GIDMaPS) Data sets. figshare. http://dx.doi.org/10.6084/m9.figshare.853801 (2014)

Q1:  Why should repositories care about data citation?  We already ask authors to cite an article that describes our resource.  Isn’t that good enough?

Q4:  What is a globally unique web-resolvable identifier and why is it important?

Q7:  Should I put a version number (i.e. semantic information) in the identifier?

Q6: How should I handle landing pages for datasets with multiple versions?

Q9:   How do I handle a data set  made of multiple parts?  Do all the parts need separate identifiers?

(e.g., misleading or inaccurate analyses and analyses aimed at unfairly discrediting or undermining the original publication)

Furthermore, a mechanism will be needed to fund the data-preparation activities necessary for data sharing in such a way as to protect confidentiality and ensure data integrity

minimum of 2 years after publication

The ICMJE proposal may also lead some investigators to delay publishing their primary trial results to allow time to prepare several secondary manuscripts. Delay or failure to publish the primary results of trials is already a substantial problem.3

Once the investigators who have conducted the trial no longer have exclusive access to the data, they will effectively be competing with people who have not contributed to the substantial efforts and often years of work required to conduct the trial.

A key motivation for investigators to conduct RCTs is the ability to publish not only the primary trial report, but also major secondary articles based on the trial data. The original investigators almost always intend to undertake additional analyses of the data and explore new hypotheses.

Adequate incentives for researchers to invest the substantial time and effort required to conduct RCTs and to publish the results in a timely fashion are important.

occasional new discoveries

The costs associated with preparing data for sharing can and should be built into the grants, cooperative agreements, and contracts that researchers negotiate with trial sponsors; in other words, expenses associated with administering data-sharing protocols must be treated as a standard, necessary aspect of the costs of carrying out a clinical trial.

new open layer

40 scholarly

the result is a standard way to enable comments and conversations in context, in a way that is not bound to a specific platform, publisher, or encoding, and in a way that makes it simple and intuitive for people to talk to people and learn from each other, and build vibrant communities that extend human knowledge.

While 'free' in the commercial sense, they silo our interactions in exact opposition to the principles on which the Web was founded, namely the democratization of knowledge.

This Workshop aims at becoming a forum to discuss ideas and advancements towards the revision of current scientific communication practices in order to support Open Science, introduce novel evaluation schemes, and enable reproducibility

irst International Workshop on Reproducible Open ScienceHannover, Germany, 9th of September 2016

obsolete brain of rodent

more constructively for creative acts rather than consumptive ones

"It could then develop a global consensus eventually and hopefully could serve as a standard for research in the future."

reventing bio-piracy raises challenges of its own. For instance Brazil has been writing new rules for taking samples outside the country. But there are reports that's leaving some researchers in limbo as they wait for the rules to be finalized.

Earlier this year some researchers from Brazil — the epicenter of the outbreak — did the public-spirited thing by uploading Zika virus genome sequences they had produced to an online public database. Soon after, scientists from Slovenia used that data in a paper they published in the New England Journal of Medicine without sharing credit.

"If you want really to have a rapid public health response you have to make sure that that data is available as soon as it's known," says Heymann. "And that means in the country."

But in fact, there were plenty of other researchers going into West Africa to do genetic sequencing. A lot of those scientists were simply waiting on prestigious journals to publish their findings.

Sometimes it's because they want to publish their results – and medical journals prefer exclusives

They don't coordinate with people fighting the epidemic on the ground — don't even share their discoveries for months, if ever.

Critics call them "parachute researchers": Scientists from wealthy nations who swoop in when a puzzling disease breaks out in a developing country.

Post the resulting data online T

By honest I don't mean that you only tell what's true. But you make clear the entire situation. You make clear all the information that is required for somebody else who is intelligent to make up their mind.Richard Feynman

Notably, too, many did their landmark work in places that some might regard as off the beaten path of science (Alicante, Spain; France’s Ministry of Defense; Danisco’s corporate labs; and Vilnius, Lithuania). And, their seminal papers were often rejected by leading journals—appearing only after considerable delay and in less prominent venues. These observations may not be a coincidence: the settings may have afforded greater freedom to pursue less trendy topics but less support about how to overcome skepticism by journals and reviewers.

The history also illustrates the growing role in biology of “hypothesis-free” discovery based on big data. The discovery of the CRISPR loci, their biological function, and the tracrRNA all emerged not from wet-bench experiments but from open-ended bioinformatic exploration of large-scale, often public, genomic datasets.

Vergnaud’s efforts to publish their findings met the same resistance as Mojica’s. The paper was rejected from the Proceedings of the National Academy of Sciences, Journal of Bacteriology, Nucleic Acids Research, and Genome Research, before being published in Microbiology on March 1, 2005.

Recognizing the importance of the discovery, Mojica sent the paper to Nature. In November 2003, the journal rejected the paper without seeking external review; inexplicably, the editor claimed the key idea was already known. In January 2004, the Proceedings of the National Academy of Sciences decided that the paper lacked sufficient “novelty and importance” to justify sending it out to review. Molecular Microbiology and Nucleic Acid Research rejected the paper in turn. By now desperate and afraid of being scooped, Mojica sent the paper to Journal of Molecular Evolution. After 12 more months of review and revision, the paper reporting CRISPR’s likely function finally appeared on February 1, 2005 (Mojica et al., 2005xIntervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements. Mojica, F.J.M., Díez-Villaseñor, C., García-Martínez, J., and Soria, E. J. Mol. Evol. 2005; 60: 174–182Crossref | PubMed | Scopus (402)See all ReferencesMojica et al., 2005).

PIs were ranked by this aggregate score and split into quintiles.

Relative Citation Ratio (RCR): A new metric that uses citation rates to measure influence at the article level

scientific productivity

and/or peer networks,

either because they measure only the average performance of a group of papers (Vinkler, 2003), or because the article of interest is measured against a control group that includes widely varying areas of science

However, high-Impact-Factor journals (JIF ≥ 28) only account for 11% of papers that have an RCR of 3 or above.

, influential publications can be found in virtually all journals

Though each of the above mentioned methods of quantitation has strengths, accompanying weaknesses limit their utility.

n recent years, decision-makers have increasingly turned to numerical approaches such as counting first or corresponding author publications, using the impact factor of the journals in which those publications appear, and computing Hirsch or H-index (Hirsch, 2005).

To perform their test, they downloaded data from old fMRI studies—specifically, information from 499 resting volunteers who were being scanned while not thinking about anything in particular (these scans were intended for use as controls in the original papers).

Computer says: oops

Finally, we point out the key role that data sharing played in this work and its impact in the future. Although our massive empirical study depended on shared data, it is disappointing that almost none of the published studies have shared their data, neither the original data nor even the 3D statistical maps.

Bibliometrics of Cluster Inference

PLoS ONE supporters have a ready answer: start by making any core text that passes peer review for scientific validity alone open to everyone; if scientists do miss the guidance of selective peer review, then they can use recommendation tools and filters (perhaps even commercial ones) to organize the literature — but at least the costs will not be baked into pre-publication charges.

And to Eisen, the idea that research is filtered into branded journals before it is published is not a feature but a bug: a wasteful hangover from the days of print.

By rejecting papers at the peer-review stage on grounds other than scientific validity, and so guiding the papers into the most appropriate journals, publishers filter the literature and provide signals of prestige to guide readers' attention. Such guidance is essential for researchers struggling to identify which of the millions of articles published each year are worth looking at, publishers argue — and the cost includes this service.

Brian Hole, founder and director of the researcher-led Ubiquity Press in London, says that average costs are £200 (US$300). And Binfield says that PeerJ's costs are in the “low hundreds of dollars” per article.

Outsell estimates that the average per-article charge for open-access publishers in 2011 was $660.

Analysts estimate profit margins at 20–30% for the industry, so the average cost to the publisher of producing an article is likely to be around $3,500–4,000.

These charges and counter-charges have been volleyed back and forth since the open-access idea emerged in the 1990s, but because the industry's finances are largely mysterious, evidence to back up either side has been lacking. Although journal list prices have been rising faster than inflation, the prices that campus libraries actually pay to buy journals are generally hidden by the non-disclosure agreements that they sign. And the true costs that publishers incur to produce their journals are not widely known.

Nature Methods' Points of Significance column on statistics explains many key statistical and experimental design concepts.

Throughout this paper, we use the phrase ‘machine actionable’ to indicate a continuum of possible states wherein a digital object provides increasingly more detailed information to an autonomously-acting, computational data explorer.

that all research objects should be Findable, Accessible, Interoperable and Reusable (FAIR) both for machines and for people.

Free publication for all Zika Virus research BMJ wants to support the fast development of research around the Zika virus by enabling researchers to share their findings as quickly as possible.

Interrupting prolonged sitting with light-intensity walking breaks may be an effective fatigue countermeasure acutely.

Reproducibility matters

It's often said that the bacteria and other microbes in our body outnumber our own cells by about ten to one.

SPM 2016 course

School of Data works to empower civil society organizations, journalists and citizens with the skills they need to use data effectively – evidence is power!

In other words, there is still plenty ofgood science out there

often reject

Good science is in some sense a public good,

we are the first to illustrate the evolutionary logic of how, in theabsence of change, the existing incentives willnecessarilylead to the degradationof scientific practices.

publication quantity

Replication allows those labs with poor methods tobe penalized, but unlessallpublished studies are replicated several times (an idealbut implausible scenario), some labs will avoid being caught.

“money-back guarantee”

This last assumption may appear unrealisticallyharsh, but research indicates that retractions can lead to a substantial decrease incitations to researchers’ prior work

When the focus is on the production of novelresults and negative findings are difficult to publish, institutional incentives forpublication quantity select for the continued degradation of scientific practices.

Research groups would never be able to get awaywith methods for whichallhypotheses are supported

The dying lab is then removed from the population. Next,a lab is chosen to reproduce

while negative novel results never are.

Scientists are humanand will therefore respond (consciously or unconsciously) to incentives; when per-sonal success (e.g., promotion) is associated with the quality and (critically) thequantityof publications produced, it makes more sense to use finite resources togenerate as many publications as possible (p. 1037, emphasis in original). Sec-ond, researchers may be trying to do their best, but selection processes rewardmisunderstandings and poor methods.

The first highly-citedexhortation to increase statistical power was published by Cohen in 1962, as a re-action to the alarmingly low power of most psychology studies at the time

The population benefitsfrom high power more than individual scientists do. Science does not possess an“invisible hand” mechanism through which the naked self-interest of individualsnecessarily brings about a collectively optimal result.

Statisti-cal power refers to the probability that a statistical test will correctly reject the nullhypothesis when it is false, given information about sample size, effect size, andlikely rates of false positives

On the other hand, low-poweredexperiments are substantially easier to perform when studying human or othermammals, particularly in cases where the total subject pool is small, the experi-ment requires expensive equipment, or data must be collected longitudinally

a dead Atlantic Salmon exhibiting neural responses toemotional stimuli

Moreover, even firmly discredited research is often citedby scholars unaware of the discreditation

For example, until recently, theJournal of Personality and Social Psychologyrefusedto publish failed replications of novel studies it had previously published

Incentives to increase one’sh-index may also encourage researchers to engage in high-risk hy-pothesizing, particularly on “hot” research topics, because they can increase their citation count bybeing corrected

This is often summarized more pithily as “when a measure becomesa target, it ceases to be a good measure.”

n the years between 1974 and 2014, the fre-quency of the words “innovative,” ”groundbreaking,” and “novel” in PubMed ab-stracts increased by 2500% or mor

ewly hired biologists now have almost twice asmany publications as they did ten years ago (22 in 2013 vs. 12.5 in 2005).

We show that the persistence of poor research practice can be explained as theresult of the natural selection of bad science.

We assume that all agents have the utmost integrity. They never cheat. In-stead, research methodology varies and evolves due to its consequences on hiringand retention, primarily through successful publication.

over 50 years of reviews of low statistical power

IF [impact factor]

Improving the quality of research requires change at the institutionallevel.

THE NATURAL SELECTION OF BAD SCIENCE

Let’s say all large publishers suddenly refused anyone any access to any of their copyrighted materials at 9am tomorrow morning — what would they be replaced with?

Give data generators proper attribution & credit from those who use their data

Scientific fields oftenwork well in situations where we can measure howwell a project is doing. In the case of processorswe know their function and we can know if ouralgorithms discover it.

As we reviewed in this paper, we may evenwant to be careful about the conclusions about themodules that neuroscience has drawn so far, afterall, much of our insights come from small datasets,with analysis methods that make questionable as-sumptions.

However, we cannot write off the failure of thesemethods on the processor simply because processorsare different from neural systems.

However,in the case of the processor we know its functionand structure and our results stayed well short ofwhat we would call a satisfying understanding

This again highlights howhard it is to derive functional insights from activitydata using standard measures.

Arguably this approach ismore justified for the nervous system because brainareas are more strongly modular.

This finding of course is grossly misleading. Thetransistors are not specific to any one behavior orgame but rather implement simple functions, likefull adders.

Wemight thus conclude they are uniquely responsiblefor the game – perhaps there is a Donkey Kongtransistor or a Space Invaders transistor.

We will finally discuss how neuroscience can worktowards techniques that will make real progress atmoving us closer to a satisfying understanding ofcomputation, in the chip, and in our brains

In this paper, much as in systems neuroscience,we consider the quest to gain an understandingof how circuit elements give rise to computation.

What constitutes an understanding of a system?Lazbnick’s original paper argued that understand-ing was achieved when one could “fix” a brokenimplementation

mportantly, the processor allows us to ask “do wereally understand this system?”

The human brain has hundreds ofdifferent types of neurons

label regions

We arguethat the analysis of this simple system implies thatwe should be far more humble at interpreting re-sults from neural data analysis.

Publishers are also getting involved: around a dozen journals this year began asking their authors to use unique identifiers for their reagents as part of a push by the Resource Identification Initiative.

NIFMOL:nifext_5242

insulating myelin sheaths

Broca’s Region: Linking Human Brain Functional Connectivity Data and Nonhuman Primate Tracing Anatomy Studies

When we start to see all the protocol, prespecified hypotheses, and raw data available for review, along with full disclosure of methods and analyses and what, if anything, changed along the course of experiments, be it at the bench or in clinical trials, we’ll have made substantive progress.

Use of blockchain technology has recently been shown to provide an immutable ledger of every step in a clinical research protocol, and this could easily be adapted to basic and experimental model science. All participants in the peer-to-peer research network have access to all of the time stamped, continuously updated data. It is essentially tamper proof since any change, such as to the prespecified data analysis, would have to be made in every computer (typically thousands) within the distributed network. All of the data would be in a findable, machine readable format. More importantly, with every step of a research path digitised and shared, we have a platform well suited for rapid, independent verification of methods and results.

What is missing is the deep commitment—across academia and the life science industry—for open science and open data.

It’s not just money that buys replication; time and experience can come into play.

Thus, the proposal from a senior executive at Merck to set up a “clawback” model, whereby an academic institution would refund its financial payment if the basic science or preclinical results prove to be irreproducible, is ironic.

Anterior cingulate area

anterior cingulate cortex

anterior cingulate gyrus

6879.23333051054

(MSH)

nlx_anat_1008012

You can find the agenda for this year’s conference here and a summary of abstracts here

400 public annotations

ETH Spam-Filter

not humanities

They are proposing to have univocally identified and persistent landing pages that provide human-readable (namely HTML) and machine-readable (e.g., JSON) basic information on the dataset. Repositories are expected to start experiencing with these approaches.

For example, repositories contribute to supply credit to data contributors, however they have not yet developed any micro attribution oriented facility aiming at highlighting who did what (Allen et al., 2014).

Repositories tend to support OAI-PMH (cf. Tab. 8) yet they should reinforce the range of protocols and facilities they offer for programmatically accessing their content, e.g., by exposing their content in formats other than HTML like Schema.org (Guha et al., 2016) and Linked Data (Bizer et al., 2009), by supporting standard protocols like OpenSearch and SRU (Denenberg, 2009). Services implemented by third-party providers largely rely on metadata, thus the quality of metadata severely impacts also the implementation of this facility e.g., Rousidis et al. (2014).

Thus trustworthiness is actually an induced property acquired by repositories during their lifetime.

The License Agreement between 3TU.Datacentrum and data submitters explicitly states that the repository (i) “shall ensure, to the best of its ability and resources, that the deposited dataset will remain legible and accessible”; (ii) “shall, as far as possible, preserve the dataset unchanged in its original format, taking account of current technology and the costs of implementation”; and (iii) “has the right to modify the format of the dataset if this is necessary in order to facilitate the digital sustainability, distribution or re-use of the dataset”.

or preservation, beside storing the data in multiple copies the selected repositories tend to use format migration practices.

Repositories might also support post-publication validation by providing mechanisms for end-users to provide both the repository and the data providers with concrete and documented feedback resulting from (re-)using or attempting to (re-)use the dataset.

Trying to assess the general ‘quality’ of a dataset is hopeless; consider instead whether the dataset is suited to a particular use.

it is almost impossible to envisage a certification scheme that guarantees that the datasets published by a repository are “scientifically sound”.

A similar approach is offered by Dryad that records the number of downloads and makes it possible to browse the “most popular” ones, in terms of downloads. 3TU.Datacentrum only publishes aggregated statistics on downloaded datasets as a sort of validation of the centrum itself.

checksum

No control is made for attesting “scientific soundness” of the dataset, e.g., scientific validity, accuracy or completeness.

Actually, it has been observed that the use of the term “peer review” for datasets is causing some false expectations (Parsons and Fox, 2013; Candela et al., 2015).

Rather than billing researchers, new payment methods can be envisaged including private- or public-sector grants and partnerships with journal publishers.

the reduction of the publishing costs,

Dryad is the only repository in our sample that always requires submitters to pay a charge independently of the files size. All the other ones offer at least a minimum storage space where users can publish free of charge.

For the sake of re-usability, it is very useful to be able to add further metadata values and documentation as the data exploitation progresses.

Unfortunately, a data paper is of use only for humans and does not provide any support for automatic consumption.

none of them explicitly requires that the deposited datasets be associated with a data paper.