Greenland has melted before, and as the climate warms, it will melt again — this time leading to what scientists warn could be 20 to 25 feet of sea-level rise.

During one of the warm periods within the last 1.1 million years, the center, not just the edges, of Greenland’s massive ice sheet melted away, new research has found, giving way to a dry and barren “tundra landscape” that was home to various insects and plant life. The findings were shared in a new paper published today in the Proceedings of the National Academy of Sciences. 

When the ice sheet initially melted, there were lower levels of greenhouse gasses in the atmosphere than there are today. Now with more carbon dioxide in the atmosphere, scientists say Greenland’s ice sheet is more susceptible to melting than previously thought. 

“Greenland has been around for 2.7 million years with its ice,” said Paul Bierman, a scientist at the University of Vermont who co-led the new study. “Now there’s some evidence that this ice sheet is fragile.”

The researchers have been studying materials from beneath the hood of the Greenland ice sheet, the largest in the Northern hemisphere, since 2014. They examined sediment from the bottom of an ice core — dubbed GISP2 — extracted from two miles below the surface at the center of the ice sheet nearly 30 years ago. 

The 1-ounce sample of sediment was filled with clues of Greenland’s past. Tiny little black specks, when put under the microscope, revealed an insect eye, an Arctic poppy seed, parts of an Arctic willow, and tiny bits of soil fungus and spike moss — what Bierman referred to as a “frozen ecosystem underneath the ice.”

According to researchers, the fossils provide “direct confirmation” that 90% of the ice sheet was once gone.

“Finding these fossils in the center of the ice sheet is unambiguous evidence that Greenland’s ice has disappeared [in the past],” said Bierman. “And once you lose the center of the ice sheet, you’ve lost it all.”

    
  
Advertisement

                                             

The findings supports what’s called the “fragile Greenland” hypothesis: that nature, outside of human influence, has caused the ice sheet to melt at least once since it formed, Bierman said. 

At 656,000 square miles, the Greenland ice sheet currently covers around 80% of the island territory. To put that into perspective, it’s about three times the size of Texas.

NASA, which has mapped Greenland’s ice loss, says the sheet has “rapidly declined in the last several years,” prompting the global sea level to rise around 0.03 inches per year. Greenland’s melting ice mass is now the No. 1 driver of sea level rise,  according to Bierman.

“In the early years of the climate warming, it was mountain glaciers that were doing most of the melting and adding water to the ocean,” he said. “Now it’s Greenland.”

While it could be a few thousand years before the entire Greenland mass melts, Bierman said, the consequences would be dire: hundreds of millions of people could lose their homes and businesses. Places we hold near and dear to our hearts would be lost.

“As I like to say when people ask me, why does it matter? I say think about your favorite beach. And then imagine your favorite beach with 25 feet of water on it,” Bierman said.

During , some more details about Monument Valley 3 emerged. Developer Ustwo Games says the builds on its predecessors in several ways, especially in terms of the gameplay, art style and story.

Perhaps the biggest change in Monument Valley 3 is the introduction of sailing. You’ll be able to move the boat in almost any direction and, as such, “the game is no longer restricted to geometric spaces,” game director Jennifer Estaris said during a call with reporters. Along with unfolding cubes that shift new protagonist Noor onto different planes, Monument Valley 3 is in large part about “deconstructing what we know,” Estaris said.

Naturally, those changes allow for fresh puzzles and visuals as players navigate impossible-looking, MC Escher-style architectural spaces. Aligned with that, Ustwo has evolved the art style with softer and more abstract shapes (you are going to be navigating rivers after all). The result is something that looks very familiar, but like a hazy memory. Meanwhile, the soundtrack from an 18-piece orchestra might just stir up some emotions.

While the original game was about forgiveness and its sequel was a coming-of-age tale, Monument Valley 3 focuses on the themes of hope, togetherness and resilience. Noor is a lighthouse keeper’s apprentice who is tasked with searching for a new source of power before light fades away from the world forever. It’s the “most ambitious story” to date for a Monument Valley game, according to Ustwo marketing manager Jamie Wotton.

All three Monument Valley games are standalone titles with their own stories. That means you’ll be able to glide right into Monument Valley 3 without playing the previous two entries. But there’s plenty of time to check those out first (or revisit them). You won’t have to pay extra for those mobile gaming classics if you have a Netflix subscription. There are no ads or in-app purchases in the company’s games.

Monument Valley is available to Netflix subscribers on iOS and Android starting today, and its sequel will arrive on the service on October 29. Monument Valley 3 will then debut as a Netflix mobile exclusive on December 10.

Netflix revealed some other mobile gaming news during Geeked Week. Its  (which includes special weapons and ranked modes) will drop on September 24. Subscribers will soon be able to play and on their mobile devices at no extra cost — the Netflix version of the latter will offer cross-play between iOS and Android.

Meanwhile, Netflix at Squid Game: Unleashed before that multiplayer tie-in game arrives later this year. Last but not least, there was a first look at Blood Line: A Rebel Moon Game, based on Zack Snyder’s Rebel Moon movies. The Netflix-exclusive co-op action game from developer Super Evil Megacorp will debut in 2025.

Twenty years ago on an Indonesian island, scientists discovered fossils of an early human species that stood at about 3 1/2 feet tall, earning them the nickname “hobbits.”

Now a new study suggests ancestors of the hobbits were even slightly shorter.

“We did not expect that we would find smaller individuals from such an old site,” said Yousuke Kaifu, co-author of the study, which was published on Tuesday in the journal Nature. 

The original hobbit fossils date back to between 60,000 and 100,000 years ago. The new fossils were excavated at a site called Mata Menge, about 45 miles from the cave where the first hobbit remains were uncovered. The fossils were found on the top of a ribbon-shaped, pebbly sandstone layer in a small stream. They included exceptionally small teeth that possibly came from two individuals, researchers said. 

In 2016, researchers suspected the earlier relatives could be shorter than the hobbits after studying a jawbone and teeth collected from the new site. Further analysis of a tiny arm bone fragment and teeth suggests the ancestors were a mere 2.4 inches shorter and existed 700,000 years ago.

“They’ve convincingly shown that these were very small individuals,” said Dean Falk, an evolutionary anthropologist at Florida State University who was not involved with the research.

Researchers have debated how the hobbits — named Homo floresiensis after the remote Indonesian island of Flores — evolved to be so small and where they fall in the human evolutionary story. They’re thought to be among the last early human species to go extinct.

Scientists don’t yet know whether the hobbits shrank from an earlier, taller human species called Homo erectus that lived in the area, or from an even more primitive human predecessor. More research — and fossils — are needed to pin down the hobbits’ place in human evolution, said Matt Tocheri, an anthropologist at Canada’s Lakehead University.

“This question remains unanswered and will continue to be a focus of research for some time to come,” Tocheri, who was not involved with the research, said in an email.

Robotics startup 1X Technologies has developed a new generative model that can make it much more efficient to train robotics systems in simulation. The model, which the company announced in a new blog post, addresses one of the important challenges of robotics, which is learning “world models” that can predict how the world changes in response to a robot’s actions.

Given the costs and risks of training robots directly in physical environments, roboticists usually use simulated environments to train their control models before deploying them in the real world. However, the differences between the simulation and the physical environment cause challenges. 

“Robicists typically hand-author scenes that are a ‘digital twin’ of the real world and use rigid body simulators like Mujoco, Bullet, Isaac to simulate their dynamics,” Eric Jang, VP of AI at 1X Technologies, told VentureBeat. “However, the digital twin may have physics and geometric inaccuracies that lead to training on one environment and deploying on a different one, which causes the ‘sim2real gap.’ For example, the door model you download from the Internet is unlikely to have the same spring stiffness in the handle as the actual door you are testing the robot on.”

Generative world models

To bridge this gap, 1X’s new model learns to simulate the real world by being trained on raw sensor data collected directly from the robots. By viewing thousands of hours of video and actuator data collected from the company’s own robots, the model can look at the current observation of the world and predict what will happen if the robot takes certain actions.

The data was collected from EVE humanoid robots doing diverse mobile manipulation tasks in homes and offices and interacting with people. 

“We collected all of the data at our various 1X offices, and have a team of Android Operators who help with annotating and filtering the data,” Jang said. “By learning a simulator directly from the real data, the dynamics should more closely match the real world as the amount of interaction data increases.”

The learned world model is especially useful for simulating object interactions. The videos shared by the company show the model successfully predicting video sequences where the robot grasps boxes. The model can also predict “non-trivial object interactions like rigid bodies, effects of dropping objects, partial observability, deformable objects (curtains, laundry), and articulated objects (doors, drawers, curtains, chairs),” according to 1X. 

Some of the videos show the model simulating complex long-horizon tasks with deformable objects such as folding shirts. The model also simulates the dynamics of the environment, such as how to avoid obstacles and keep a safe distance from people.

Challenges of generative models

Changes to the environment will remain a challenge. Like all simulators, the generative model will need to be updated as the environments where the robot operates change. The researchers believe that the way the model learns to simulate the world will make it easier to update it.

“The generative model itself might have a sim2real gap if its training data is stale,” Jang said. “But the idea is that because it is a completely learned simulator, feeding fresh data from the real world will fix the model without requiring hand-tuning a physics simulator.”

1X’s new system is inspired by innovations such as OpenAI Sora and Runway, which have shown that with the right training data and techniques, generative models can learn some kind of world model and remain consistent through time.

However, while those models are designed to generate videos from text, 1X’s new model is part of a trend of generative systems that can react to actions during the generation phase. For example, researchers at Google recently used a similar technique to train a generative model that could simulate the game DOOM. Interactive generative models can open up numerous possibilities for training robotics control models and reinforcement learning systems. 

However, some of the challenges inherent to generative models are still evident in the system presented by 1X. Since the model is not powered by an explicitly defined world simulator, it can sometimes generate unrealistic situations. In the examples shared by 1X, the model sometimes fails to predict that an object will fall down if it is left hanging in the air. In other cases, an object might disappear from one frame to another. Dealing with these challenges still requires extensive efforts.

One solution is to continue gathering more data and training better models. “We’ve seen dramatic progress in generative video modeling over the last couple of years, and results like OpenAI Sora suggest that scaling data and compute can go quite far,” Jang said.

At the same time, 1X is encouraging the community to get involved in the effort by releasing its models and weights. The company will also be launching competitions to improve the models with monetary prizes going to the winners. 

“We’re actively investigating multiple methods for world modeling and video generation,” Jang said.