After making up a false claim about a nonexistent study done by the AAAS, the AI software admitted that it made a mistake and then apologized.
The post Did ChatGPT Just Lie To Me? appeared first on The Scholarly Kitchen.
After making up a false claim about a nonexistent study done by the AAAS, the AI software admitted that it made a mistake and then apologized.
The post Did ChatGPT Just Lie To Me? appeared first on The Scholarly Kitchen.
by ZBW MediaTalk-Team
Ever since Elon Musk, holding a sink in his arms (“Let that sink in!”), entered the Twitter headquarters in San Francisco at the end of October, a sense of dark foreboding has been spreading in the online world. The richest man in the world had orchestrated a hostile takeover of the short message service: It is rumoured to have cost him 44 billion US dollars to turn his hobby into a new enterprise, which he can add to his business empire (Tesla, SpaceX, SolarCity, Neuralink and others).
The billionaire had previously assured the world that he is a “free speech absolutist”. His plan was now to make Twitter into a place of uncensored freedom of speech. Those who were sanctioned and blocked for violating the community rules, would sooner or later receive a general absolution and be able to return to the platform. Even Donald Trump – former president of the United States and co-instigator of the most spectacular attempted coup in the USA to date – would have the red carpet rolled out for him.
Now Twitter has never been a cosy refuge of mutual understanding, consideration and the cultured exchange of arguments. Twitter has polarised opinions for years. But as the increase in social division has continued, particularly in the west, hate and toxicity have been constantly increasing on the platform. They are expressed in threats, open racism, discrimination, fake news, doxing and cyber-bullying. More than a few German politicians have therefore recently pulled the plug and turned their backs on the network.
How Twitter will develop in future years is anyone’s guess. However, on the evidence of the few days since Elon Musk has been at the helm, it doesn’t look good. The new CEO appears to be nervously driven, almost erratic. His first act after taking the wheel was to fire the moderating powers within the company, thereupon to bark contradictory commands to the remaining workforce. In the meantime, Twitter Inc. has neither a press department nor a data protection officer, causing the data protection officers of German companies and organisations to break out in a collective sweat, because the operation of Twitter accounts under consideration of GDPR aspects can only be legally justified with a great deal of good will.
Ministries, authorities but also the science sector is now facing a dilemma. There is a strong moral obligation to pack up, shut down the account that you have been nurturing and maintaining for many years and bid farewell, softly but firmly, to Twitter. On the other hand, there is an understandable fear of loss of reach: How can politics stay in touch with the public? How can universities, museums and libraries fulfil their public mandate if, at the same time, they leave their online communities?
It is questions like these that, since the dimming of Twitter, have led to one name in particular being floated around: “Mastodon”. At the moment it’s individuals in particular, who are looking for a new home – and the short messaging service alternative seems to have a certain appeal to members of the science community especially.
Much has been written in recent days about this actually not-so-very-new platform. Started in 2016 by German software developer Eugen Rochko, it is a distributed micro-blogging service that lies completely in the hands of the community, thanks to its open source code. In contrast to Twitter, Mastodon is not a centrally organised entity but a network that is created from hubs “instances”. Every instance can function autonomously or alternatively stretch its arms out to the big network where it then becomes part of the large Fediverses that is home these days not only to social networks but also to video streaming services, image sharing services and the like. Theoretically, every imaginable service and every kind of content can be added to the Fediverse using compatible open source communication protocols – the possibilities are boundless!
Theoretically, at least.
Although the developments regarding the Twitter takeover are to be evaluated critically, they were – at the same time – a collective wake-up call for openness in the digital sphere. The idea of decentralised systems that are in the hands of the communities – such as for scientific exchange and scholarly communication – is closely aligned to the wish of many people for more openness in science. There is no gatekeeper; there are no paywalls, no evolved, incomprehensible hierarchies; just the self-organisation of the community.
ZBW MediaTalk succumbed to the charm of Mastodon at a quite early stage. In 2019 we set up the account for the blog; a few months ago we really got going and since then we have been posting content regularly from the library and Open Science world.
And it’s working.
But after several months of operation, maybe it’s time to do a stocktake – not a performance evaluation, though; it’s definitely too early for that. But a summary of the experiences we have made to date. Because naturally even this much lauded network (perhaps occasionally praised with too much uncritical euphoria) is not entirely free of problems. Let’s refer to them as unusually deep puddles that lurk out of sight, and that Mastodon newbies can easily put their feet into. Because they do exist.
At that time we decided to create our account on the Openbiblio Instance. Purely theoretically though, we could have decided to use any one of the dozens of official and even hundreds of unofficial instances. Or to operate our own server. So why Openbiblio? This instance has been operated by the Berlin State Library (SBB) since 2019, and we therefore know the team behind it. There is a data protection statement, server rules and thanks to the maintenance by the SBB IT department, one can assume that the accessibility of the server is relatively reliable. All this is not necessarily a matter of course. As a result of its decentralised nature, Mastodon and the Fediverse in general have been born with structural weaknesses that have still not been ironed out.
1. Data protection
2. Data security
Next keyword: data security. This too depends completely on the knowhow and commitment of the server administrator. It doesn’t take much to bring a Mastodon instance to life. But it doesn’t take much to destroy it again either. The founder of the Social.Bonn server found this out in the year 2017. When trying to install an update on his instance, the whole system crashed: all postings and all the accounts that had been previously set up were irretrievably lost. There was no backup.
Do the administrators of the chosen instance handle it with care? Do they install critical fixes to the code in a timely manner? Do they even install updates at all? Can they guarantee regular data security? From the outside, these questions can almost never be answered, which means that choosing an instance is reduced to a game of chance. The hint that you can change your instance at any time is no help here, because when would be the right time to do this? However much the world mistrusts the major commercial platforms: no-one seriously worries about a complete loss of data there.
A third point of criticism concerns the climate – the social discourse on the platform. How can it be ensured that the instance is a place of civilised discourse? Mastodon is by default equipped with features that allow the members to report offensive or criminal content. But how and whether the administrators react to the reports is initially left solely up to them. The Fediverse does not have a common canon of values for content evaluation; there are no generally-valid community guidelines and no overriding committee that members can call on for clarity if no action is taken or suspicions are false. What mobbing is, what fake news is, where offensiveness stops and open hatred begins – all this is decided by the administrators of the respective server, initially under their own steam. Sometimes their rules are laid down specifically; sometimes not. Factors such as the size of an instance and the resources available can also make content moderation more difficult. The large commercial networks rely on artificial intelligence and outsourced moderation teams to fish out evil, dirty and forbidden content from the timelines. How can just one person take on this task round the clock if they are maintaining an instance with thousands of members? And the issue of toxicity is only one element of the supervision: we haven’t even mentioned how copyright-protected content is handled (German).
Data protection, data security and moderation – these are the three critical weak points that you need to bear in mind with Mastodon, when choosing an instance. There is always only an approximation of security (and at this point, thanks again to the SBB in Berlin), but no guarantees. If you want to play it safe, you logically have to rely on self-hosted instances.
Operating your own instances as an alternative to using the services of the major commercial players sounds like the promised land in a science environment that is becoming increasingly more open, transparent and independent. This is also true in the light of current efforts to have the operation of Open Science infrastructures completely in the hands of scientific communities (scholarly-owned) or at least under their control /scholarly-led). But in order for this plan to become a reality, institutions must cooperate more closely, come to agreements, and develop a common vision of what such a network could look like and the values it could reflect. And the time is now. Consolidation, clear responsibilities and transparency are required to minimise the three structural weak points. One idea could be to establish a consortium, within which several scientific institutions can join forces, either on an institutional or target group-specific basis, in order to jointly operate an instance that is secure for everyone. The fact that Mastodon is an open source project means that there is even the opportunity to actively push the development of the network forward or promote it in another way.
Alternatively or additionally, the development of a certification process is a possibility for existing and new instances such as those in the scientific sector. A joint criteria catalogue has been defined for this purpose – compliance with it offers registered users a certain degree of security. Are there specific contact persons? Is data protection maintained? Are data pools backed up regularly? Does moderation take place, and if yes, on the basis of which rules? If there was simply a seal, a formal certification, then outsiders would have many of their questions answered. Even today, timid attempts at an initial regulation have been made: For example, the official Mastodon website currently only lists those servers, who fulfil certain criteria, although this tends to concern merely rudimentary rules.
These are just a few suggestions. There are sure to be a few clever ideas out there on this topic that could help to make Mastodon a viable alternative to Twitter – or much more, perhaps. One thing is certain: the momentum to start thinking about it has arrived right now.
You may also be interested in:
Featured Image: Wikipedia, Creative Commons Attribution-ShareAlike 4.0.
The post Self-organised network: does Mastodon have what it takes to become the “scholarly-owned social network”? first appeared on ZBW MediaTalk.
Another “mixed bag” post from us — Is it time to leave Twitter? How can we incentivize journals and authors to take up open science practices? What is “involution” and is DEIA the solution?
The post Smorgasbord: Twitter v. Mastodon; Incentivizing Open Science; DEI v. Involution appeared first on The Scholarly Kitchen.
A new type of post from us today, offering a smorgasbord of opinions on topics including the ongoing Twitter/Elon Musk saga, just what “equitable access” to the literature means, the ongoing lack of experimental controls in one area of bibliometric analysis, and whether journals are more like a gate or a sewer.
The post Smorgasbord: A Better Metaphor for Publishing, Twitter/Musk, Equitable Access, and Those Vexing OACA Experimental Controls appeared first on The Scholarly Kitchen.
by Claudia Sittner
Profiles of public institutions on commercial social networks have long been a source of unease for data protectionists throughout Germany. But until 2016, there was a lack of equivalent alternatives. Mastodon could be one such alternative. The decentralised network by software developer Eugen Roschko is Open Source with the source code being freely available on Github. The decentralised nature of Mastodon is what makes it so attractive and a good choice from an Open Science perspective.
Unlike platforms such as Facebook or Twitter, it is non-commercial, free of advertising and run by volunteers, who can also be organisations, on so-called instances. The instances are nodes, i.e. servers, creating a decentralised network: The Fediverse – a cross between “Federated” and “Universe”. The Fediverse is the generic concept for a network of federated systems, which implements a specific protocol (ActivityPub) – like Mastodon.
The network is not financed by advertising revenue, which in practice almost always means the collection and use of personal data. So it can be operated in compliance with data protection laws. For data protectionists, however, the decentralisation is the big bonus. It is considered more data-protection compliant because not all data is collected at a central point. Of course, commercial providers can also host an instance, but that would only be one of many. Many Mastodon nodes also have their own data protection statements. However, there are no contracts for Data Processing Agreements. The choice of an instance is, therefore, ultimately a matter of trust.
The increased compatibility with data protection led, among other things, to some German authorities (German) and data protection officers, such as the data protection officer of Baden-Württemberg, Stefan Brink (German), setting up a profile there.
On Mastodon, users can create accounts in two steps. In the first step, they choose a suitable instance. These differ in terms of the community that has settled there, the usage guidelines, the number of users, the language and the tone. The latter is defined in the usage rules of the instance. Anyone who does not abide by them can be excluded by the admins. Since the individual instances have far fewer accounts than other social networks, expulsion on Mastodon can actually happen quite quickly. It is even possible to exclude entire instances from the Fediverse if, for example, there are no rules of use or the users do not adhere to them and serious cross-account violations occur too frequently.
There are currently around 3,800 Mastodon instances and 5.2 million users (July 2022). Filter systems or blog posts such as this one (German) can help you find the right instance. When we created the MediaTalk account on Mastodon in 2018 – back then in the tail of the Cambridge Analytica scandal – we chose the OpenBiblio instance. It is run by the Staatsbibliothek zu Berlin (state library). OpenBiblio is constantly growing and currently hosts almost 400 profiles. Around 29,000 contributions have been posted here so far. In the meantime, the instance has become well established for the German-language library scene. In the second step, you can create the user account. The account URL also includes the name of the instance, in our case: https://openbiblio.social/web/@ZBW_MediaTalk.
As can be seen from this example, the name of the instance can already say something about its background, thematic focus and the interests of its community. But don’t worry, choosing an instance is not a life decision: if you don’t feel at home, it is possible to switch. Followers can be taken along, some other things too, but unfortunately not postings.
To stay with our example: Just because our account is hosted on the OpenBiblio instance does not mean that we can only follow others on that instance and only see their posts, because all instances are linked to the Fediverse. Communication is often compared to e-mail: Even if I have a Googlemail account, I can write mails to users of a Hotmail account and receive mails from them. The practical thing is that with a Mastodon account, you can also follow users with profiles on other Fediverse services like Pixelfed, which is about photos.
So we can follow all other accounts and vice versa. Nothing stands in the way of writing your own posts, which are called “toots” on Mastodon. A toot consists of a maximum of 500 characters. As usual, pictures, videos or similar attachments can be sent along. What is called a “retweet” on Twitter is a “boost” on Mastodon. Instead of hearts or likes, users award stars.
After the fuss about Elon Musk’s tweet about buying Twitter, we decided to fill our profile on Mastodon with life. Since we are doing this in test mode for the time being, we are mirroring our tweets there. We currently use the command line tool t2m for this. It’s a bit on the old side, but it’s Open Source and can be easily operated on your own server. There are also online tools for this form of mirroring, but their use in compliance with data protection regulations is questionable. For the reasons mentioned, we are now active on Mastodon, for the time being. But we are confident that we will be there permanently. In what form, after a certain test phase, is still open. So if you are no longer happy on Twitter, maybe because of data protection concerns, you won’t miss anything if you follow us on Mastodon. We would be happy to welcome you there!
This text has been translated from German.
Read more about Mastodon:
Read more on MediaTalk:
Claudia Sittner studied journalism and languages in Hamburg and London. She was a long time lecturer at the ZBW publication Wirtschaftsdienst – a journal for economic policy, and is now the managing editor of the blog ZBW MediaTalk. She is also a freelance travel blogger (German), speaker and author. She can also be found on LinkedIn, Twitter and Xing.
Portrait: Claudia Sittner©
Featured Image: Mastodon press kit
The post Briefly Noted: ZBW MediaTalk in Test Mode on Mastodon first appeared on ZBW MediaTalk.
Twitter does not increase citations, a reanalysis of author data shows. Did the authors p-hack their data?
The post Desperately Seeking (Statistical) Significance appeared first on The Scholarly Kitchen.
When a reputable journal refuses to get involved with a questionable paper, science looks less like a self-correcting enterprise and more like a way to amass media attention.
The post Fill in the Blank Leads to More Citations appeared first on The Scholarly Kitchen.
PLOS ONE is organizing a Twitter chat on citizen science methodologies on 2nd April- see details below Citizen science (CS) encompasses a broad range of research methodologies that involve public participation for data collection, transcription
PLOS ONE is organizing a Twitter chat on citizen science methodologies on 2nd April- see details below
Citizen science (CS) encompasses a broad range of research methodologies that involve public participation for data collection, transcription or analysis. Applications of CS have been found in many disciplines, but ecology has consistently been at the forefront. While some CS-based ecological monitoring schemes- such as the UK Butterfly Monitoring Scheme, established in 1976- have been running for decades, the popularity of CS has grown rapidly in more recent years. A wide range of projects based on CS methodologies are now being undertaken around the world, at local, national and international scales. The value of volunteer participation in activities ranging from transect-based species monitoring (Wepprich et al., 2019) and collection of biological specimens for lab-based analysis (Larson et al., 2020; Rasmussen et al., 2020) to crowdsourcing of creative thinking for study design (Can et al., 2017), has been repeatedly demonstrated. Studies have also highlighted the particular utility of CS methodologies in supporting long-term ecological monitoring in resource-limited contexts, including in economically developing countries (Gouraguine et al., 2019). Meanwhile, examples of the real-world impact of CS research are abundant, both in specific ecological interventions and the wider political discourse. For instance, the influential UK State of Nature 2019 report, likely to be a key source of evidence for future environmental legislation, cites the outcomes of a wealth of CS projects.
With the expansion of CS research, there is lively debate about how to maximise rigor and reproducibility in different types of CS methodologies. One of the crucial aspects of a successful CS study is an appropriately designed protocol, which features a realistic degree of complexity and accounts for the specific challenges of handling CS-derived data. An example of this is provided by a recent comprehensive report of the design, launch and assessment of the UK National Plant Monitoring Scheme (Pescott et al., 2019). Pre-testing of protocols prior to project launch can provide confidence in the robustness of the study design. When designing a CS study, it is also important to understand volunteer motivation and ensure that this is appropriately matched with the nature of the task to be performed (Lyons & Zhang, 2019). Some CS studies utilize narrower demographic groups to meet the required level of motivation and understanding, such as amateur naturalists (Hallmann et al., 2017) or students who are following a course in a related topic (Chiovitti et al., 2019). Depending on the type of study, researchers may also plan to support CS volunteers with training or technological aids, increasingly in the form of mobile apps (Ožana et al., 2019; Appenfeller et al., 2020).
A certain amount of error, either random or systematic, is likely to be introduced by the collection of data by CS volunteers, and study designs must account for this. The level of error can be reduced by allowing volunteers to provide clarifying metadata or to register uncertainty (Torre et al., 2019), or using incentives to reduce sampling bias (Callaghan et al., 2019), but researchers should also ensure that they have means to assess the accuracy of contributed data (Falk et al., 2019; Gibson et al., 2019). Much ecological research is based on large public databases of volunteer-contributed records of species distributions, phenological events and other observational data (e.g. Siljamo et al., 2020). There is an active discussion in the ecological research community about how to maximize the reliability and utility of such data (Ball-Damerow et al., 2019).
The particular considerations that have to be made in the design, execution and evaluation of CS studies has led to calls for dedicated standards and guidelines for CS research. Of course, any such tools must strike the balance between promoting appropriate levels of standardization and allowing the flexibility required for applications of CS methodologies across diverse settings and research questions. Whilst some progress has been made towards this goal, maintaining an open and constructive dialogue among CS practitioners and other stakeholders remains critical to ensure that researchers, volunteers and society are able to realize the full potential of CS.
To foster discussion of these important issues, PLOS ONE (@plosone) will be moderating a Twitter chat on citizen science methodologies on Thursday 2nd April starting at 4pm BST (8am PDT, 11am EDT, 5pm CET). This is a chance for the CS community to share perspectives, experiences and suggestions for best practice. We’ll aim to cover the following questions (and more!):
You can take part by using the hashtag #citscichat– we hope to see you there!
Appenfeller LR, Lloyd S, Szendrei Z (2020) Citizen science improves our understanding of the impact of soil management on wild pollinator abundance in agroecosystems. PLoS ONE 15(3): e0230007. https://doi.org/10.1371/journal.pone.0230007
Ball-Damerow JE, Brenskelle L, Barve N, Soltis PS, Sierwald P, Bieler R, et al. (2019) Research applications of primary biodiversity databases in the digital age. PLoS ONE 14(9): e0215794. https://doi.org/10.1371/journal.pone.0215794
Callaghan CT, Rowley JJL, Cornwell WK, Poore AGB, Major RE (2019) Improving big citizen science data: Moving beyond haphazard sampling. PLoS Biol 17(6): e3000357. https://doi.org/10.1371/journal.pbio.3000357
Can ÖE, D’Cruze N, Balaskas M, Macdonald DW (2017) Scientific crowdsourcing in wildlife research and conservation: Tigers (Panthera tigris) as a case study. PLoS Biol 15(3): e2001001. https://doi.org/10.1371/journal.pbio.2001001
Chiovitti A, Thorpe F, Gorman C, Cuxson JL, Robevska G, Szwed C, et al. (2019) A citizen science model for implementing statewide educational DNA barcoding. PLoS ONE 14(1): e0208604. https://doi.org/10.1371/journal.pone.0208604
Falk S, Foster G, Comont R, Conroy J, Bostock H, Salisbury A, et al. (2019) Evaluating the ability of citizen scientists to identify bumblebee (Bombus) species. PLoS ONE 14(6): e0218614. https://doi.org/10.1371/journal.pone.0218614
Gibson KJ, Streich MK, Topping TS, Stunz GW (2019) Utility of citizen science data: A case study in land-based shark fishing. PLoS ONE 14(12): e0226782. https://doi.org/10.1371/journal.pone.0226782
Gouraguine A, Moranta J, Ruiz-Frau A, Hinz H, Reñones O, Ferse SCA, et al. (2019) Citizen science in data and resource-limited areas: A tool to detect long-term ecosystem changes. PLoS ONE 14(1): e0210007. https://doi.org/10.1371/journal.pone.0210007
Hallmann CA, Sorg M, Jongejans E, Siepel H, Hofland N, Schwan H, et al. (2017) More than 75 percent decline over 27 years in total flying insect biomass in protected areas. PLoS ONE 12(10): e0185809. https://doi.org/10.1371/journal.pone.0185809
Larson RN, Brown JL, Karels T, Riley SPD (2020) Effects of urbanization on resource use and individual specialization in coyotes (Canis latrans) in southern California. PLoS ONE 15(2): e0228881. https://doi.org/10.1371/journal.pone.0228881
Lyons E, Zhang L (2019) Trade-offs in motivating volunteer effort: Experimental evidence on voluntary contributions to science. PLoS ONE 14(11): e0224946. https://doi.org/10.1371/journal.pone.0224946
Ožana S, Burda M, Hykel M, Malina M, Prášek M, Bárta D, et al. (2019) Dragonfly Hunter CZ: Mobile application for biological species recognition in citizen science. PLoS ONE 14(1): e0210370. https://doi.org/10.1371/journal.pone.0210370
Pescott OL, Walker KJ, Harris F, New H, Cheffings CM, Newton N, et al. (2019) The design, launch and assessment of a new volunteer-based plant monitoring scheme for the United Kingdom. PLoS ONE 14(4): e0215891. https://doi.org/10.1371/journal.pone.0215891
Rasmussen SL, Nielsen JL, Jones OR, Berg TB, Pertoldi C (2020) Genetic structure of the European hedgehog (Erinaceus europaeus) in Denmark. PLoS ONE 15(1): e0227205. https://doi.org/10.1371/journal.pone.0227205
Siljamo P, Ashbrook K, Comont RF, Skjøth CA (2020) Do atmospheric events explain the arrival of an invasive ladybird (Harmonia axyridis) in the UK? PLoS ONE 15(1): e0219335. https://doi.org/10.1371/journal.pone.0219335
Torre M, Nakayama S, Tolbert TJ, Porfiri M (2019) Producing knowledge by admitting ignorance: Enhancing data quality through an “I don’t know” option in citizen science. PLoS ONE 14(2): e0211907. https://doi.org/10.1371/journal.pone.0211907
Wepprich T, Adrion JR, Ries L, Wiedmann J, Haddad NM (2019) Butterfly abundance declines over 20 years of systematic monitoring in Ohio, USA. PLoS ONE 14(7): e0216270. https://doi.org/10.1371/journal.pone.0216270
All images used under Pixabay licence.
The post Maximizing the rigor and reproducibility of citizen science in ecological research appeared first on EveryONE.
The post “Low T” and Prescription Testosterone: Public Viewing of the Science Does Matter appeared first on EveryONE.