One week until holidays, so quite a busy schedule in the lab, trying to finish as much as possible before the break. This week we have some interesting links on image manipulation, climate change, remote sensing by using fibers, and a cool application of Artificial Intelligence. Let’s start:
Nothing to see here
A piece of news that’s been there for a couple weeks, but I did not have time to read in detail. Some scientists have discovered digital image tampering on several high impact papers related to Alzheimer disease. We are talking about dozens of publications from a small group of authors, done with public money, and on such a hot topic that it is possible that they might have mislead hundreds of researchers, wasting millions in funding, into a dead-end.
The moment science became a competition and scientists started fighting for funding, stuff like this was bound to happen. While this is not the first time I see accusations of people manipulating images to publish cool papers (in high impact factor journals, obviously), it is probably the most impactful research misbehaviour I have ever seen. Of course, people will always try to fool others to achieve their goals, but peer review should be the mechanism we use to protect ourselves from this. However, the system right now works by exploiting scientists, who review papers in very tight time schedules and without getting paid (i.e. either in their free time or during they working hours, both of which are not cool). You cannot expect reviewers to perform digital forensics on each image they see on a draft, more so when there is usually no access to the original files, and they only get to see a pdf file. The thing even gets worse when computational techniques are involved: people rarely provide full datasets, the code is obscure and not always shared, and fields such as Machine Learning require high-end equipment just to replicate some results.
Faked Beta-Amyloid Data. What Does It Mean?, on Science
Detecting earthquakes with your internet cables
Really cool piece on how scientists are taking advantage of telecom infrastructure in the bottom of our oceans to detect Earthquakes. Optical fibers have been used many times in industry to monitor humidity, gas leaks, and temperature changes, but it is nice to see the same ideas applied in such a massive scale and to this kind of problem.
A few stories related to climate change this week. First, the old problem of first-world countries looking at poor ones from a high moral ground. You cannot expect poor people to not use their natural resources because they are not green when you have been exploiting the whole world during centuries, pillaging and ravaging any piece of land until you became rich. The only solution I see is to invest in those places and make it so they do not need to destroy their environment to survive, period.
In March, Ève Bazaiba, Congo’s minister of environment, told The New York Times that officials were mulling going ahead. “Should we protect peatland because it’s a carbon sink or should we dig for oil for our economy?” she said.
The second one was quite interesting. It tells the story of several scientists who made changes in their careers in order to fight climate change. Some of them were quite big, some of them not so drastic, but all share the main objective: improving Earth’s health.
Given her role as a wildlife ecologist at the University of Idaho in Moscow, it might seem that Gilbert was already well placed to have a positive impact on climate change. But the slow, incremental pace of academia, and the difficulty of getting policymakers to act on her findings, left her feeling that she was not making as much of a difference as she’d hoped. “I’ve been studying how wildlife responds to environmental change to inform conservation planning for 15 years now, researching and publishing and waiting for something to happen and then having it not happen, even when I’ve worked closely with wildlife and land-management agencies,” she says. “The system just isn’t designed to respond to the urgent challenges we’re facing,” she says.
Next one is something we see every year, but each time it happens earlier. Earlier this week, we consumed the amount of natural resources that our planet provides during a whole year. And there are still 22 weeks until 2023. The trend is also not very optimistic: we have been increasing our consumption nonstop for the past +50 years, and there are no signs of slowing down in the near future…
And last one is an incredible leak on how power companies do whatever they can (without even stopping to think about legality) to keep their benefits and hinder the adoption of cleaner forms of energy.
The CEO of the biggest power company in the US had a problem. A Democratic state senator was proposing a law that could cut into Florida Power & Light’s (FPL) profits. Landlords would be able to sell cheap rooftop solar power directly to their tenants – bypassing FPL and its monopoly on electricity. “I want you to make his life a living hell … seriously,” FPL’s CEO Eric Silagy wrote in a 2019 email to two of his vice-presidents about state Senator José Javier Rodríguez, who proposed the legislation. Within minutes, one of them forwarded the directive to the CEO of Matrix, LLC, a powerful but little-known political consulting firm that has operated behind the scenes in at least eight states. Rodríguez was ousted from office in the next election. Matrix employees spent heavily on political advertisements for a candidate with the same last name as Rodríguez, who split the vote. That candidate later admitted he was bribed to run.
Congo to Auction Land to Oil Companies: ‘Our Priority Is Not to Save the Planet’, on The New York Times
Alarm as Earth hits ‘Overshoot Day’ Thursday: NGOs, on Phys.org
‘Earth Overshoot Day’ comes earlier every year, on France24
Leaked: US power companies secretly spending millions to protect profits and fight clean energy, on The Guardian
I always bring news regarding miss-practices of AI, so it is also good to write when the field provides cool stuff. In this case, AlphaFold’s results in predicting the structure of ~200 million proteins is a feat worth mentioning. Also, more good news: even though the feat was achieved by the private sector, the information will be publicly available for anyone to use. I like to think of this as potentially having the same impact as the periodic table of the elements had in the XIX and XX centuries (or even more).
DeepMind has predicted the structure of almost every protein known to science, on the MIT Technology Review
And that’s it for the week. Stay safe!
Featured image: Vitellogenin’s protein structure — DeepMind