It’s possible that I shall make an ass of myself. But in that case one can always get out of it with a little dialectic. I have, of course, so worded my proposition as to be right either way (K.Marx, Letter to F.Engels on the Indian Mutiny)
The possible orbit of Planet Nine. Credit: CalTech/R. Hurt (IPAC)
There are eight known planets in the solar system (ever since Pluto was booted from the club), but for a while, there has been some evidence that there might be one more. A hypothetical Planet 9 lurking on the outer edge of our solar system. So far, this world has eluded discovery, but a new study has pinned down where it should be.
The evidence for Planet 9 comes from its gravitational pull on other bodies. If the planet exists, its gravity will affect the orbits of other planets. So if something seems to be tugging on a planet, just do a bit of math to find the source. This is how Neptune was discovered, when John Couch Adams and Urbain Le Verrier noticed independently that Uranus seemed to be tugged by an unseen planet.
In the case of Planet 9, we don't have any gravitational effect on a planet. What we do see is an odd clustering of small icy bodies in the outer solar system known as Kuiper belt objects (KBOs). If there were no planet beyond the Kuiper belt, you would expect the orbits of KBOs to be randomly oriented within the orbital plane of the solar system. But instead, we see lots of KBO orbits are clustered in the same orientation. It's possible that this is just due to random chance, but that isn't likely.
Back in 2016, the authors looked at the statistical distribution of KBOs and concluded the clustering was caused by an undetected outer planet. Based on their calculations, this world has a mass of five Earths and is about 10 times more distant from the sun than Neptune. The paper even calculated a broad region of the sky where the planet might be. But searches turned up nothing. This led some to conclude the planet doesn't exist. Orbital oddness doesn't prove a planet exists. Just ask Planet Vulcan. Others went so far as to argue Planet 9 does exist, but we can't see it because it's a primordial black hole.
This new study reexamines the original work in light of some of the criticism it received. One big criticism is that outer solar system bodies are difficult to find, so we look for them where it's convenient. The clustering effect we see could just be due to biased data. Taking observational bias into effect, the authors find the clustering is still statistically unusual. There's only a 0.4% chance of it being a fluke. When they recalculated the likely orbit of Planet 9, they were able to better localize where to look.
One interesting aspect of the study is that the newly calculated orbit puts Planet 9 closer to the sun than originally thought. This is odd, because if it is closer then we should have already found it. The authors argue that observations thus far have ruled out the closest options for Planet 9, which helps narrow down its possible location even further. If the planet exists, it should be detectable by the Vera Rubin Observatory in the near future.
This study isn't conclusive, and many astronomers still argue that Planet 9 doesn't exist. But this study makes it clear that we won't have to argue about it for much longer. Either it will be discovered soon, or observations will rule it out as an explanation for the KBO clustering effect.
Machines and robots undoubtedly make life easier. They carry out jobs with precision and speed, and, unlike humans, they do not require breaks as they are never tired.
As a result, companies are looking to use them more and more in their manufacturing processes to improve productivity and remove dirty, dangerous, and dull tasks.
However, there are still so many tasks in the working environment that require human dexterity, adaptability, and flexibility.
Human-robot collaboration is an exciting opportunity for future manufacturing since it combines the best of both worlds.
The relationship requires close interaction between humans and robots, which could highly profit from anticipating a collaborative partner's next action.
Ph.D. student Achim Buerkle and a team of researchers from the Intelligent Automation Centre at Loughborough University have published promising results for 'training' robots to detect arm movement intention before humans articulate the movements in the Robotics and Computer-Integrated Manufacturing journal.
"A robot's speed and torque need to be co-ordinated well because it can pose a serious threat to human health and safety", said Achim.
"Ideally, for effective teamwork, the human and robot would 'understand' each other, which is difficult due to both being quite different and 'speaking' different languages."
"We propose to give the robot the ability to 'read' its human partners intentions."
The researchers looked to achieve this by interfacing the frontal lobe activity of the human brain.
Every movement performed by the human body is analyzed and evaluated in the brain prior to its execution. Measuring this signal can help to communicate an 'intention to move' to a robot.
However, brains are highly complex organs, and detecting the pre-movement signal is challenging.
The Loughborough University researchers tackled this challenge by training an AI system to recognize the pre-movement patterns from an electroencephalogram (EEG) – a piece of technology that allows human brain activity to be recorded.
Their latest paper reports the findings of a test carried out with eight participants.
The participants had to sit in front of a computer that randomly generated a letter from A-Z on the screen and press the key on the keyboard that matched the letter. The AI system had to predict which arm the participants would move from the EEG data and this intention was confirmed by motion sensors.
The experimental data shows that the AI system can detect when a human is about to move an arm up to 513 milliseconds (ms) before they move, and on average, around 300ms prior to actual execution.
In a simulation, the researchers tested the impact of the time advantage for a human-robot collaborative scenario.
They found they could achieve a higher productivity for the same task using the technology as opposed to without it.
The completion time for the task was 8-11% faster—even when the researchers included 'false positives', which involved the EEG wrongly communicating a person's intention to move to the robot.
Achim plans to build on this research and hopes to eventually create a system that can predict where movement is directed—for example, reaching for a screwdriver or picking a new work piece.
Of the latest findings, he said that "we hope this study will achieve two things: first, we hope this proposed technology could help towards a closer, symbiotic human-robot collaboration, which still requires a large amount of research and engineering work to be fully established."
"Secondly, we hope to communicate that rather than seeing robots and artificial intelligence/machine learning as a threat to human labor in manufacturing, it could also be seen as an opportunity to keep the human as an essential part of the factory of the future."
In a joint statement, Achim's supervisors Dr. Thomas Bamber, Dr. Niels Lohse, and Dr. Pedro Ferreira said that "there is a need to transform the nature of human work in order to create a truly sustainable world no longer dependent on strenuous physical and cognitive human labor."
"Human-Robot Collaboration (HRC) is starting to innovate factory shop-floors, however, there is still a need for more substantial collaboration between humans and robots."
"True HRC will have a transformative effect on labor productivity, job quality, and wellness and establish a more secure and sustainable labor market, whilst also overcoming physical disadvantages caused by gender, sex age, or disability."
"Achim's work using Artificial Intelligence and EEG brings us one step closer to true HRC."
More information:Achim Buerkle et al, EEG based arm movement intention recognition towards enhanced safety in symbiotic Human-Robot Collaboration,Robotics and Computer-Integrated Manufacturing(2021).DOI: 10.1016/j.rcim.2021.102137
Autocorrection, or predictive text, is a common feature of many modern tech tools, from internet searches to messaging apps and word processors. Autocorrection can be a blessing, but when the algorithm makes mistakes it can change the message in dramatic and sometimes hilarious ways.
Our research shows autocorrect errors, particularly in Excel spreadsheets, can also make a mess of gene names in genetic research. We surveyed more than 10,000 papers with Excel gene lists published between 2014 and 2020 and found more than 30% contained at least one gene name mangled by autocorrect.
This research follows our 2016 study that found around 20% of papers contained these errors, so the problem may be getting worse. We believe the lesson for researchers is clear: it's past time to stop using Excel and learn to use more powerful software.
Excel makes incorrect assumptions
Spreadsheets apply predictive text to guess what type of data the user wants. If you type in a phone number starting with zero, it will recognize it as a numeric value and remove the leading zero. If you type "=8/2," the result will appear as "4," but if you type "8/2" it will be recognized as a date.
With scientific data, the simple act of opening a file in Excel with the default settings can corrupt the data due to autocorrection. It's possible to avoid unwanted autocorrection if cells are pre-formatted prior to pasting or importing data, but this and other data hygiene tips aren't widely practiced.
In genetics, it was recognized way back in 2004 that Excel was likely to convert about 30 human gene and protein names to dates. These names were things like MARCH1, SEPT1, Oct-4, jun, and so on.
Several years ago, we spotted this error in supplementary data files attached to a high impact journal article and became interested in how widespread these errors are. Our 2016 article indicated that the problem affected middle and high ranking journals at roughly equal rates. This suggested to us that researchers and journals were largely unaware of the autocorrect problem and how to avoid it.
As a result of our 2016 report, the Human Gene Name Consortium, the official body responsible for naming human genes, renamed the most problematic genes. MARCH1 and SEPT1 were changed
to MARCHF1 and SEPTIN1 respectively, and others had similar changes.
An example list of gene names in Excel.
An ongoing problem
Earlier this year we repeated our analysis. This time we expanded it to cover a wider selection of open access journals, anticipating researchers and journals would be taking steps to prevent such errors appearing in their supplementary data files.
We were shocked to find in the period 2014 to 2020 that 3,436 articles, around 31% of our sample, contained gene name errors. It seems the problem has not gone away, and is actually getting worse.
Small errors matter
Some argue these errors don't really matter, because 30 or so genes is only a small fraction of the roughly 44,000 in the entire human genome, and the errors are unlikely to overturn to conclusions of any particular genomic study.
Anyone reusing these supplementary data files will find this small set of genes missing or corrupted. This might be irritating if your research project examines the SEPT gene family, but it's just one of many gene families in existence.
We believe the errors matter because they raise questions about how these errors can sneak into scientific publications. If gene name autocorrect errors can pass peer-review undetected into published data files, what other errors might also be lurking among the thousands of data points?
In 2012, JP Morgan declared a loss of more than US$6 billion thanks to a series of trading blunders made possible by formula errors in its modeling spreadsheets. Analysis of thousands of spreadsheets at Enron Corporation, from before its spectacular downfall in 2001, show almost a quarter contained errors.
A now-infamous article by Harvard economists Carmen Reinhart and Kenneth Rogoff was used to justify austerity cuts in the aftermath of the global financial crisis, but the analysis contained a critical Excel error that led to omitting five of the 20 countries in their modeling.
In biomedical research, a mistake in preparing a sample sheet resulted in a whole set of sample labels being shifted by one position and completely changing the genomic analysis results. These results were significant because they were being used to justify the drugs patients were to receive in a subsequent clinical trial. This may be an isolated case, but we don't really know how common such errors are in research because of a lack of systematic error-finding studies.
Better tools are available
Spreadsheets are versatile and useful, but they have their limitations. Businesses have moved away from spreadsheets to specialized accounting software, and nobody in IT would use a spreadsheet to handle data when database systems such as SQL are far more robust and capable.
However, it is still common for scientists to use Excel files to share their supplementary data online. But as science becomes more data-intensive and the limitations of Excel become more apparent, it may be time for researchers to give spreadsheets the boot.
In genomics and other data-heavy sciences, scripted computer languages such as Python and R are clearly superior to spreadsheets. They offer benefits including enhanced analytical techniques, reproducibility, auditability and better management of code versions and contributions from different individuals. They may be harder to learn initially, but the benefits to better science are worth it in the long haul.
Excel is suited to small-scale data entry and lightweight analysis. Microsoft says Excel's default settings are designed to satisfy the needs of most users, most of the time.
Clearly, genomic science does not represent a common use case. Any data set larger than 100 rows is just not suitable for a spreadsheet.
Researchers in data-intensive fields (particularly in the life sciences) need better computer skills. Initiatives such as Software Carpentry offer workshops to researchers, but universities should also focus more on giving undergraduates the advanced analytical skills they will need.
Apple CEO Tim Cook attends Apple's "Ted Lasso" season two premiere event red carpet at the Pacific Design Center, in West Hollywood, California, July 15, 2021.
Apple chief executive Tim Cook has received a bonus of some $750 million, reflecting his performance at the US technology giant in his 10 years at the helm, a regulatory filing showed.
The Thursday filing with the Securities and Exchange Commission showed Cook's bonus was granted in some five million Apple shares, which were subsequently cashed out.
The bonus reflects a stock option plan implemented in 2011 when Cook took over as CEO from Steve Jobs shortly before the Apple co-founder's death.
Since then, Apple's market value has skyrocketed and the California giant's worth is estimated at more than $2.4 trillion.
The stock award included 1.1 million shares granted based on Apple's performance and 3.9 million time-based share awards, all of which were vested this month.
Cook still owns some 3.2 million Apple shares worth $483 million at today's value.
His net worth is estimated by Forbes magazine at some $1.4 billion, well below that of other Silicon Valley tech leaders.
The Alabama native held a variety of jobs at Apple before taking over from Jobs as CEO.
Under his leadership, the iPhone maker expanded into new areas of digital content and subscriptions to diversify its revenue stream.
Power networks worldwide are faced with increasing challenges. The fast rollout of distributed renewable generation (such as rooftop solar panels or community wind turbines) can lead to considerable unpredictability. The previously used "fit-and-forget" mode of operating power networks is no longer adequate, and a more active management is required. Moreover, new types of demand (such as from the rollout EV charging) can also be source of unpredictability, especially if concentrated in particular areas of the distribution grid.
Network operators are required to keep power and voltage within safe operating limits at all connection points in thenetwork, as out of bounds fluctuations can damage expensive equipment and connected devices. Hence, having good estimates of which area of the network could be at risk and require interventions (such as strengthening the network, or extra storage to smoothen fluctuations) is increasingly a key requirement.
Privacy-sensitive machine learning
Smart meter data analysis holds great promise for identifying "at risk" areas in distribution networks. Yet, using smart meter data can present significant practical constraints. In many countries and regions, the rollout of smart meters does not provide full coverage, as installation is voluntary and many customers may reject installing a smart meter at their home. Moreover, even places where there is a successful smart meter roll-out, privacy restrictions must be taken into account and, in practice, regulators considerably constrain what private data from smart meters network operators have access to.
Newly published research from the Smart Systems Group at Heriot-Watt University in Edinburgh, UK, in collaboration with Scottish Power Energy Energy Networks addresses these key challenges. Based on real data and case studies from distribution networks in Scotland, researchers have shown that deep learning neural networks can provide accurate estimates of voltage distributions in all areas of the network, even if high-granularity smart meter data is available from only a few locations, not from every consumer meter.
Dr. Maizura Mokhtar, the Data Scientist who led the work, explains that"while modern smart meters can collect high-granularity data from every household, in practice, there are computational constraints with collecting so much data, as well as privacy concerns. Our work shows that, to produce high accuracy voltage predictions across the whole network, only data from a few Key Identified Locations is needed. Furthermore, it can do so by using only current voltage data to output accurate voltage predictions. Crucially, our method does NOT require input of privacy-sensitive power data, which could be conceivably be used to infer what individual customer activity in their home."
Dr. Valentin Robu, Associate Professor and Academic PI of the project, says that "this work was part of the NCEWS (Network Constraints Early Warning System project), a collaboration between Heriot-Watt and Scottish Power Energy Networks, part funded by InnovateUK, the United Kingdom's applied research and innovation agency. The project's results greatly exceeded our expectations, and it illustrates how advanced AI techniques (in this case deep learning neural networks) can address important practical challenges emerging in modern energy systems. We were very much honored to win the IET and E&T 2019 Innovation of the Year Award for the work in this project, as well as now a top publication in the Energy and AI journal."
Professor David Flynn, the Head of the Smart Systems Group at Heriot-Watt added that "the NCEWS project showcases very well how an academia-industry collaboration can bring new thinking and expertise to the activity of UK power network operators. Artificial Intelligence and data analytics are increasingly central to addressing challenges that UK DNOs face, and will likely play a key role in decarbonising our energy systems."
More information:Maizura Mokhtar et al, Prediction of voltage distribution using deep learning and identified key smart meter locations,Energy and AI(2021).DOI: 10.1016/j.egyai.2021.100103
Elizabeth Holmes was worth an estimated $3.6 billion in 2014, according to Forbes.
A female entrepreneur whose multi-billion dollar start-up looked set to revolutionize medical testing, before it crashed and burned in a blaze of fraud claims, goes on trial next month.
When she launched the diagnostics company Theranos in 2003, at the age of just 19, the charismatic Elizabeth Holmes promised results that were faster and cheaper than traditional laboratories, running an analytical gamut on just a few drops of blood.
Political grandees like Henry Kissinger and James Mattis were drawn to the company's board; media mogul Rupert Murdoch invested his cash in what seemed to be a sure-fire winner.
Holmes was lauded as a visionary, drawing comparisons with Apple founder Steve Jobs.
But years of hype, and billions of dollars later, those promises unspooled; the miracle machines did not work.
And, say prosecutors, Holmes knew it, yet continued to lie to investors, doctors and patients so she could raise more than $700 million.
At her long delayed trial, which begins in September in San Jose, California, Holmes faces nine charges of wire fraud and two of conspiracy to commit wire fraud that could see her jailed for up to 20 years.
"In terms of how much money the government has alleged has been lost, this case is actually not the biggest health care fraud case in the past year. There are many larger health care fraud cases than the Theranos trial," says Jason Mehta, a lawyer and former prosecutor specializing in health care fraud.
"But in terms of media attention and scrutiny, it certainly ranks right up there as one of the bigger cases in the past decade."
'Real patient harm'
Jurors, who will be selected from Tuesday, could hear from former board members like Kissinger and Mattis, as well as from Murdoch.
But they may also hear from patients who were victims of the faulty tests, and received misdiagnoses of cancer, HIV or pregnancy.
Holmes' lawyers may want to avoid such testimony, but for Mehta, their experiences are key to understanding what happened.
"It's what makes these cases real. It's what makes the government's allegation not just about dollars and cents, but about real patient harm.
"For many jurors, the patient testimonial likely will be the most compelling and the most moving testimony."
Holmes, seen here with Chinese entreprenur Jack Ma and former US president Bill Clinton was an object of fascination way beyond the Bro culture of Silicon Valley start-ups.
It may also be necessary for prosecutors to prove their case; the 2018 dismantling of Theranos' servers rendered a key company database unreadable.
But for all the potential star power of the witness roster, the most eagerly awaited testimony could come from Holmes herself, if she decides to take the stand.
According to John Carreyrou, the Wall Street Journal reporter who first broke the story and has since written a book about the scandal, Holmes was a true believer in her vision of cheap blood tests—even if she knew the facts didn't really fit.
"Not to say that she didn't know along the way that she was lying at certain junctures and cutting corners, but she felt that was justifiable and the goal was noble," he told CNBC last month.
$3.6 billion
The combative Holmes was an object of fascination way beyond Silicon Valley for more than a decade.
A highly successful woman in a world dominated by macho venture capitalists and the "bro" culture of start-ups, she amassed an estimated $3.6 billion, according to Forbes in 2014.
At the time she was the youngest billionaire not to have inherited her fortune.
The high tolerance of risk that brought her immense wealth is also what has led her to battle the charges in court, and chance of a much lengthier jail term, says Carreyrou.
"Nine out of 10 or even 99 out of 100 people in her shoes would have copped a plea several years ago, but she is willing to roll the dice and take it all the way to court."
In 2018, the Securities and Exchange Commission presented the Theranos case as a lesson for Silicon Valley, a warning against the "fake it till you make it" culture.
And for biotech and healthcare start-ups, it's a shadow that's difficult to escape because of the constant comparisons in the public mind.
"I had to waste a lot of time talking about Theranos to people who weren't in the space and in particular, as a fund manager who raises funds from generalist investors... a question we often get is 'How do you avoid the next Theranos?'," said Jenny Rooke, managing director of Genoa Ventures.
"I stifle a laugh and try to be patient here. But the next Theranos is not our problem.
"Those of us who are experts can see the big delta between claims like that and what is possible, and can see the red flags around the lack of peer review and the lack of expert involvement and the lack of transparency."
The trial has been postponed several times—notably because the defendant had a child in early July. It is scheduled to begin on September 7 and last several months.
Tesla on part-automated drive system slams into police car
This Feb. 9, 2019, file photo shows a sign bearing the company logo outside a Tesla store in Cherry Creek Mall in Denver. A Tesla using its partially automated driving system slammed into a Florida Highway Patrol cruiser Saturday, Aug. 28, 2021 on an interstate near downtown Orlando and narrowly missed its driver, who had pulled over to assist a disabled vehicle. Credit: AP Photo/David Zalubowski, File
A Tesla using its partially automated driving system slammed into a Florida Highway Patrol cruiser Saturday on an interstate near downtown Orlando and narrowly missed its driver, who had pulled over to assist a disabled vehicle.
Earlier this month, the U.S. government opened a formal investigation into Tesla's Autopilot driving system after a series of similar collisions with parked emergency vehicles.
The trooper whose cruiser was hit shortly before 5 a.m. Saturday had activated his emergency lights and was on the way to the disabled vehicle when the Tesla hit the cruiser's left side and then collided with the other vehicle, highway patrol spokeswoman Lt. Kim Montes told The Orlando Sentinel.
The report said the 27-year-old man in the Tesla and the driver of the disabled vehicle suffered minor injuries and the trooper was unhurt.
Tesla did not immediately respond to an email sent to its press address.
Autopilot has frequently been misused by Tesla drivers, who have been caught driving drunk or even riding in the back seat while a car rolled down a California highway.
The electric vehicle maker uses a camera-based system, a lot of computing power, and sometimes radar to spot obstacles, determine what they are, and then decide what the vehicles should do. But researchers say it has had trouble with parked emergency vehicles and perpendicular trucks in its path.
The National Highway Traffic Safety Administration opened the Tesla probe after tallying 11 crashes since 2018 in which Teslas on autopilot or cruise control have hit vehicles where first responders have used flashing lights, flares, an illuminated arrow board or cones warning of hazards.
In those crashes, 17 people were injured and one was killed, the NHTSA said. An investigation could lead to a recall or other enforcement action.
The National Transportation Safety Board, which also has investigated Tesla crashes, has recommended that NHTSA and Tesla limit the autopilot's use to areas where it can safely operate. It also recommended that Tesla be required to improve its system to ensure drivers pay attention.
Last year the NTSB blamed Tesla, drivers and lax regulation by NHTSA for two collisions in which Teslas crashed beneath crossing tractor-trailers.
The crashes into emergency vehicles cited by NHTSA began on Jan. 22, 2018, in Culver City, California, near Los Angeles when a Tesla using autopilot struck a parked firetruck with flashing lights. No one was injured in that accident.
Crypto money gains traction in adult industry amid OnlyFans drama
There is growing interest in bitcoin and other cryptocurrencies for adult content, but some say these systems remain cumbersome.
Porn stars, sex workers and others in adult entertainment were taking a closer look at cryptocurrency payments in the wake of a series of troubles with the mainstream financial system, potentially propelling digital currencies into wider use.
The latest problem came earlier this month when OnlyFans announced it would ban sexually explicit content on the sex-friendly creator site, only to reverse course days later following a backlash.
Nonetheless, the drama could accelerate a move to cyptocurrency to allow anonymous payments to performers outside the banking system.
With stricter rules from payment processors and the recent issues with OnlyFans, "it's obvious crypto will be the solution," said British performer Adreena Winters, who is also a brand ambassador for an upcoming crypto-friendly adult content marketplace.
"Porn has frequently been the factor for new concepts taking off, be it VHS, online credit card payments and even the internet, so I don't think it's surprising that porn will be what eventually get crypto to become mainstream."
Jeff Dillon, chief development officer at Nafty, a cryptocurrency platform launched this year specifically for the adult industry, said the OnlyFans saga "has done more than any marketing we could ever paid for."
Sex leads tech
Dillon said the sex industry has paved the way for other innovations online, such an online credit card payments and instant verification, and that it may do the same for cryptocurrency if payment processors make it more difficult.
"This is going to catapult momentum for crypto and alternative payment solutions," he said.
Dominic Ford, founder of JustFor.Fans, an OnlyFans rival which accepts bitcoin, said crypto represents just a small fraction of transactions on his platform because it is more cumbersome, but suggested this could ramp up quickly if popular money transfer tools adapt.
"A cryptocurrency that works online and transcends borders seems an obvious evolution like email was the evolution of mail," said Ford.
CumRocket, a startup which created a digital coin called Cummies for adult content, announced in recent days it was accelerating work on its own content platform.
"Sex workers should have the opportunity to join a platform that won't be subject to any payment processing restrictions, something that the other OnlyFans alternatives that use fiat may be subject to in the upcoming months/year."
While bitcoin and other digital currencies have seen extreme volatility, adult operators say they can avoid those issues by using them for immediate payments without storing them.
US law and liability
OnlyFans was not the only online service to struggle with acceptance of mature content.
PornHub has been accepting cryptocurrency for its premium service "to keep current with our community's privacy and payment preferences."
Visa and Mastercard temporarily banned payments last year to sites owned by porn giant MindGeek, which owns PornHub and other sites, over reports that it was hosting non-consensual "revenge porn".
And this month, US lawmakers demanded an investigation into alleged child pornography on OnlyFans.
Ford said congressional passage of the FOSTA-SESTA law in 2018 created pressure on the adult content industry by holding online services liable for illegal content such as child exploitation or sex trafficking.
Shortly after passage of the law, the social network Tumblr banned explicit content, resulting in a precipitous drop in usage.
Crypto may be a mixed blessing, said US-based adult content creator Deon Glows, helping circumvent some of the restrictions in the banking system but also bringing in customers "seeking anonymity for unethical reasons."
"There is skepticism (on crypto) because sex workers want to make the barriers to entry as minimal as possible," she said.
"I'd like to see banking institutions and payment processors get with the times and be more liberal about the kind of businesses they allow."
Some adult operators say crypto is promising but not ready for the majority of users.
"We will be looking to implement crypto and other alternative payment mechanisms as a backup and a option to support crypto enthusiasts but certainly not as a primary source for accepting or sending funds," said a spokesperson for the British-based adult social media platform Unlockd.
Lou Kerner, a cryptocurrency investor and analyst with Quantum Economics, said the adult industry could help bring crypto to more users.
"It's hard for people who work in the industry to get bank accounts. So they've been discriminated against for many years," Kerner said.
"As the technology becomes easier to use, more in the porn industry will adopt it... Crypto is undoubtedly on its way to mainstream adoption, and the more industries that are ill served by traditional finance, the faster it will get there."
Swedish lecturer Gabriel Erikkson Sahlin, a 24-year-old trans man, regularly faces transphobic abuse on Twitch.
Night after night, video game streamer RekItRaven watches as their feed is inundated with abusive messages. Hate raided, yet again.
In recent months the phenomenon of "hate raids"—barrages of racist, sexist and homophobic abuse—has been making life increasingly unpleasant for minority users of Twitch, the world's biggest video game streaming site.
Raven, a Black 31-year-old who identifies as gender non-binary, fought back tears while describing the mental toll of logging into a site designed for entertainment.
"It just gets hard," said the parent-of-two, who declined to reveal their real name over fears for their security.
"I'm being hated on for my skin color, or my sexual preferences, when I don't have control over that."
Twitch is more than a source of fun for Raven: it's their job. The Virginia-based horror games player holds affiliate status, under which prolific and widely followed streamers get paid.
Sick of racial slurs and messages referring to the Ku Klux Klan, Raven started a Twitter hashtag, #TwitchDoBetter.
The hashtag has become a magnet for complaints over the past month, largely from female, non-white and LGBTQ players, that Twitch is failing to stop internet trolls running amok—all while taking 50 percent of streamers' earnings.
Attack of the bots
Launched in 2011 and bought by Amazon three years later, Twitch counts more than 30 million visitors per day, most of whom tune in to watch other people play video games with entertaining commentary.
But that doesn't mean it can't be used for serious purposes.
Twitch streamer RekItRaven, a Black 31-year-old who identifies as gender non-binary, gets "hated on" over skin color and sexual preferences.
Swedish lecturer Gabriel Erikkson Sahlin logs in under the username BabblingGoat to play "The Sims" and "Dragon Age".
The 24-year-old trans man gently answers questions in the live chat about gender identity—including from anxious parents whose children have recently come out—"while falling down ledges in games and trying not to die", he said, laughing.
He is frustrated that his efforts to create something positive are being disrupted, with alarming regularity, by transphobic abuse.
The hate raids vary in scale: they can involve a handful of people posting hateful messages, or hundreds.
People also program bots to post endless offensive spam, sometimes in the form of "gore raids"—volleys of ultra-violent images.
Inventive trolls
Under increasing pressure, Twitch this month announced that new measures to prevent hate raids, including "account verification improvements", would be introduced later this year.
In the meantime, players say there has been no let-up in the abuse.
"The hate raids have not slowed down whatsoever. They only seem to be getting worse," said Chonki, a Jewish-Chinese New Yorker whose stream was inundated with anti-Semitic messages and images of swastikas.
Players have various tools at their disposal to try to filter abuse and block bullies.
But hate raid victims say the trolls use "leet" hacker slang—deliberately misspelling words—to carry on using banned terms, or they embed abusive words in images to avoid detection.
The stream of Chonki, a Jewish-Chinese New Yorker, was inundated with anti-Semitic messages and images of swastikas.
Amusement, boredom, revenge
Mark Griffiths, director of the International Gaming Unit at Britain's Nottingham Trent University, said determined trolls would "always find ways around" the tools designed to stop them.
His research over the past 25 years has found people usually troll for three main reasons: amusement, boredom or revenge.
The "perceived anonymity" offered by pseudonyms on platforms like Twitch—even though users may ultimately be identifiable—makes people feel empowered to do and say things they wouldn't normally, he added.
Streamers have proposed various ways in which Twitch could better identify perpetrators and keep them off the site, including requiring two-factor authentication.
Twitch declined to comment on a list of Raven's suggestions passed on by AFP.
Until the site comes up with more permanent solutions, players are doing their best to support each other.
But with no end to the hate raids in sight, marginalized streamers say Twitch's appeal is increasingly outweighed by the psychological burden.
Erikkson Sahlin is determined to stay because his educational streams have "been able to help so many people".
"But it's very, very taxing," he said. "This morning I was like, 'Do I really want to stream tonight? There's a 99 percent chance I'm going to get hate raided'."
For Chonki and Raven, both of whom rely on Twitch for their livelihood, there is extra pressure to keep logging in, despite the unhappiness it causes them.
"Twitch is taking 50 percent of my income—of all streamers' income—and they can't even protect us from hate raids," Chonki said.
