In the final two minutes of a 2 hour and 40 minute simulation of the Titanic sinking, its creators decided to introduce the sounds of screams. The viral, real-time video that has racked up more than 6 million views in the past week is almost otherwise devoid of voices. And you never see a human for its entirety.

The simulation is “mesmerizing” or “eerie” or “slow, beautiful and unsettling,” depending on which headline you ask.

It’s also a way for its creators to raise money for a video game they hope to make — one that, like the real-time video, they hope will be read as a respectful tribute. “This game will allow the player to see this heroism first hand and will stand as a testament to the bravery and self-sacrifice of that fateful night,” the game’s developers explain in their promotional material.

In the final version of this project, the strange silence of much of the simulation that went viral last week will give way to a murder mystery, layered on top of the real-life tragedy of the Titanic. Although the game has no timeline for release, its creators hope it will be available in VR on Oculus Rift, “to make this game as mind-blowing as possible.”

Not all simulated tragedies are ones that the Internet accepts, in the way it has the Titanic video. Last fall, a group of French college students made the ill-advised [8:46] for VR, a simulation of the Sept. 11, 2001, attacks. The player experienced the attacks from the perspective of someone inside the North Tower; its creators told reporters that they wanted to “remind that 9/11 was, for the victims, first and foremost a workday like every other workday.”  They drew heavily from journalistic re-creations of the attacks, and interviewed a survivor, they said.

The headlines for [8:46] used very different words: “Horrifying and distasteful.” With the exception of the trailer and play-through videos still available on YouTube, it looks like, at this point, the 9/11 simulator has been shamed off of most of its original Internet footprint.

Despite these very different reactions, the two attempts at simulation have things in common: Both creators made games based on tragedies that resulted in massive loss of life. And both have said that their simulations were heavily researched, and are experiences intended to honor the victims of these events, to allow the people who didn’t experience them to understand them better.

As journalist Ava Kofman pointed out in a recent presentation at Theorizing the Web in New York, there’s been a trend toward calling VR an “empathy machine,” a label that Kofman then questions. Some are taking VR’s perceived potential for empathy and running with it: A growing group of developers has created simulations of harrowing experiences, like sexual assault or the refugee crisis, with the intention of triggering empathy in the player — even if the reality of what’s going on is more complicated. [8:46] might be a stark recent example of how the “empathy machine” can go wrong; the viral success of the Titanic simulation, which isn’t even in VR yet, is a good example of why developers may believe it will work.

It’s worth noting that although viral tragedy simulations, driven into the public eye either by outrage or awe, are definitely a virtual reality right now, they’re not a problem the Internet created.

Stereographs — or image pairs that, when viewed together, created a 3D effect — were popular in the late 19th and early 20th century. It was relatively affordable, and became a well-known source of entertainment. An 1859 Atlantic article, marveling at stereographs, asserted that “by this instrument that effect is so heightened as to produce an appearance of reality which cheats the senses with its seeming truth.”

But as is happening with simulations today, the stereograph wasn’t just used for light entertainment and travel photos. The Library of Congress’s archives also contain stereographs of disasters, like the 1906 earthquake in San Francisco.

“Ruined but not disheartened, his safe, his business souvenir, San Francisco, April 1906,” from the Library of Congress.

Liked that? Try these: