The United States faces a terrifying perspective: the destruction of several parts of its national scientific infrastructure. Open data sets are closed, including decades of climate monitoring at the National Oceanic and Atmospheric Administration (NOAA) and health information for Centers for Disease Control and Prevention (CDC) – a particularly worrying loss of neuroscience. The publications financed by the federal government containing certain keywords, such as “women” and “bias”, were prohibited. And, according to some observers, PubMed could soon be flooded With non-scientific anti-vaccination “papers”. Even certain independent charitable organizations show disturbing signs of capitulating the new and anti-scientific requirements of the new administration, perhaps for fear of losing their tax exemption status: the Howard Hughes Medical Institute finance aimed at making science more inclusive, Just like the Chan Zuckerberg initiative.
International science has been to depend on this infrastructure in danger, and the training effects of the implosion of scientific institutions in the United States could be devastating worldwide.
He is tempting in the face of these desperately desperately desperate damage to the way things were before. But not only is it ineffective, because the new American administration will be in power for at least four years (and is certainly more possible), but it is a failure to learn the lesson of this moment: we must stop counting on scientific infrastructure provided by a nation or organization. Any unique point of failure makes science fragile. Instead, we need several organizations in as many countries as possible, collectively providing access to data and services that overlap, so that the loss of one or more of them does not prevent us not to do science. There are short -term steps that we can take to protect ourselves, but we must also start the work of strengthening long -term resilience.
H
Ow have things finished like that? Part of the problem is that the construction and maintenance of all kinds of infrastructure is difficult and costly, and, for most scientists, this does not correspond to our skills or inclinations. Naturally, we prefer to let someone else do it so that we can focus on science. But that puts us in a vulnerable situation, and this is not the first time that we have felt the consequences of this.
A recent dramatic example is twitter. Many scientists make enormous efforts in the construction of networks to communicate with colleagues and the general public. But all this work and the value of these networks were lost when many scientists felt forced to leave after taking control of Elon Musk of the platform (now X). The process of rebuild on Bluesky is underway, but it will take years and can never reach the same critical mass. Even if the transition is successful, the same can happen to Bluesky in a few years.
These are not only social media: we see the same problem happening in some of the main science institutions. Historically, scientists owned and directed their own journals, but they were convinced selling to publishers in the 1950s to the 1960s. These publishers only merged in a few companies and have spectacularly increased publication costs, giving rise to some of the most beneficial margins in the world. More worrying, companies have removed control of scientists. Even when we are sure that a document is fraudulent, it often takes years before being retracted, if never ever, because the retractions are simply not good for business. Most of the major scientific publishers are based in Europe, but a huge percentage of its income comes from the United States if the Trump administration required that these publishers block the unrealized publications of the United States, would they hold and fight- they?
Even pre -printed servers such as arxiv And biorxiv Can be vulnerable because they are recorded in the United States if it remains firm, their accommodation suppliers can be a point of vulnerability. In 2024, Arxiv stopped reflecting his data and went to a cloud supplier based in San Francisco. And what about other essential servers? How disruptive would it be to Github began to remove benchmarks, or if Google Scholar was starting to hide certain articles in response to requests from the United States government? After all, Google has already been very happy to Rename the whole seas To respond to Trump’s whims.
S
o What can we do? In the short term, we must try to use a more diverse range of services located in several countries. Versions of the European Union (as Zenodo instead of Osf for the deposit of data and manuscripts, or EuropePMC Instead of Central Pubmed) are a good start because the EU is an organization with several countries. We must not stop there, however, because even the EU has been vulnerable to political and corporate influence – as, for example, when they were ready to demand the scientific edition in free access in 2007 but watered At the last minute after an intense lobbying of industry.
We must also start to integrate resistant interference services in our daily workflows. Publish is a post-publication revision service widely used by scientific “detectives” to share evidence of fraud. With a wider adoption and ideally recorded outside the United States, it could make scientific literature resisting the type of influence that the US government is trying. You do not need to rely on the newspaper if the commentary on the scientific community is open and visible for everyone on each article. Elife Offers an intermediate approach, making a publication decision before the exam, then rendering the process of examination and revision fully transparent. On social networks, Mastodon is managed and managed by the community (for example, neuromatch.social is a server managed by neuroscientists) and is robust by design. It is slightly more difficult to use, which naturally makes many people, but it improves quickly.
In the longer term, we must build robust infrastructure in a more rigorous and systematic way. Part of this can be carried out by the developers of existing services, adding capacities to protect themselves from the possibility of themselves being compromised. For example, the platform Equal research includes an “poisoned pill” agreement Among its shareholders to make it more difficult to sell. Other approaches may include facilitation and encouragement of sharing and openly license of your code and data so that new services can be easily turned in the event of compromise. Biorxiv, for example, publishes an easily downloadable flow of all its data.
In the end, we must start building new multi-country organizations, designed from the start to be managed by the community and resilients. A coalition of university libraries could be ideally placed to direct this effort, because they already have a wide experience for building and supporting access to scientific data and publications. Simply imagine a world in which universities support each other by building scientific infrastructure based on free sharing, giving each country, from the poorest to the richest and the surest and surest of scientific data that can change the world to best. It is not fanciful. It can happen, but we have to make the effort, because no one will do for us.