I think this is an error in grammar.

I believe what they're saying there is that with a higher-power RF amplifier that is purpose-built to operate in a vacuum, they could test in even higher vacuum than they were able to during this test. The section is Summary and Forward Work and I don't think they're saying that they did not test in a vacuum, but that their ability to test in a vacuum was limited and could be improved in future work. 5x10^-6 torr is not quite "vacuum of outer space"; it's a high vacuum, but not quite interstellar-space vacuum.

It's probably #2. The paper, as presented at the 50th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, is available for purchase. I happened to have a spare $25 and a burning curiosity. The full paper isn't available on the NASA site, only the abstract can be gotten there for free. If you wanna read the details, you have to pay for 'em.

Anyhow, here's the relevant bit from the paper: "Two roughing pumps provide the vacuum required to lower the environment to approximately 10 Torr in less than 30 minutes. Then, two high-speed turbo pumps are used to complete the evacuation to 5x10E-6 Torr, which requires a few additional days. During this final evacuation, a large strip heater (mounted around most of the circumference of the cylindrical chamber) is used to heat the chamber interior sufficiently to emancipate volatile substances that typically coat the chamber interior walls whenever the chamber is at ambient pressure with the chamber door open. During test run data takes at vacuum, the turbo pumps continue to run to maintain the hard vacuum environment."

I'm not a physicist, but the paper is still an absolutely fascinating read, and contains a number of color photos of the test apparatus, the device itself, etc. The amount of detail they went into for the experiment is really impressive; seismically isolating the test chamber, using liquid metal (galinstan) electrical contacts to eliminate any forces due to a mechanical coupling to a wire, compensating for the magnetic field that is created by passing electricity through the device, and so on. This is NASA we're talking about here, the guys that do ROCKET SCIENCE. The idea that they wouldn't test this device in a vacuum is laughable.

Something spooky is going on inside this device, and I hope it doesn't take us too long to figure out what is really happening.

The crash transcript would be some guy...PC LOAD LETTER?! What the fuck does that--radio silence------

... to what Tor already leaks, is the previous hop from which the exit traffic came, and possibly meta data on other tunnels relayed by (but not terminated at) the node. If the relayed connection is SSL/TLS encrypted, that encryption is end-to-end from the original client to the server; sniffing some exit-node memory does not help you there. If the related connection is in the plain, then, well, then sniffing the exit node's memory does not tell you any more than you already knew by looking at its plain-text traffic.

Now, Heartbleed is not completely harmless here: You may, if you're very lucky, be able to sniff the previous node name, but as Tor tunnels are longer than that, that does not help you much. Plus, tunnels endpoints tend to change every couple of minutes, making the cross section even smaller. Also, you may now be in a position to sniff data from nodes whose ISP network you do not control, allowing you to do network-wide attacks. That may in fact be the biggest problem.

Stephen Thaler's creativity machine is proof of the potential of machine creativity.

...somehow did not make it into the summary:

Strandbeest is Dutch for Beach Animal.