5 ms·
Spying is a big problem in Eve. At the most basic level a spy is able to take screenshots and copy/paste text to send back to the entity they're spying for. By
by lagadu 9y ago
Spying is a big problem in Eve. At the most basic level a spy is able to take screenshots and copy/paste text to send back to the entity they're spying for. By watermarking both text and forum backgrounds the data effectively becomes tainted in the sense that the screenshot/text will have unique characteristics that allow the original poster to identify who copied the data in the first place.
The whole flow would be:
- X is a player in alliance FOO but is actually an agent planted there by alliance BAR.
- Y is a player in alliance BAR but is actually an agent planted there by alliance FOO (or any other).
- X copies intel from FOO's forums and sends it back to the people in charge of the spy program in BAR.
- That intel gets shared with key personnel in alliance BAR, so they can take action based on the gathered intelligence; unknown to BAR, among that personnel is Y
- Y sees that intel and sends it back to their leadership in alliance FOO
- The people at alliance FOO identify the unique data on that intel and track it down to X and proceed to kick him for being a spy.
As someone who is involved with alliance leadership stuffs: This may sound convoluted but it's really bread-and-butter level stuff in Eve, it can get significantly weirder. For example this article discusses whitespace character-based fingerprinting: that's amateur level, like described elsewhere in this thread.
To answer your question: counter-intelligence. If your method for tagging data is known and easy to replicate, such as watermarking your forum userID in screenshots, the people at alliance BAR can edit the screenshot (or forge a new one) where they insert the userID of an innocent person in alliance FOO in the gathered intel, this way when Y grabs the data for sending back to FOO, they'll be unknowingly sending evidence that incriminates a loyal member.
edit: I noticed I didn't address your last question. Signing data is just regular cryptographic signing yeah, in the above example to prevent tampering you'd insert the hash from the userID plus a secret salt for example. You just need some way to prevent the hostiles from incriminating someone else.
- ThrustVectoring 9y ago> The people at alliance FOO identity the unique data on that intel and track it down to X and proceed to kick him for being a spy. That, or deliberately feed misinformation to the spy.
- Fnoord 9y agoScreenshots made in World of Warcraft contain a watermark which contains the account number of the player, as well as other info [1] [2]. This is used to find/combat cheaters (e.g. botters). As for spying, I'm pretty sure this is a problem in high end raiding in WoW. But due to the nature of WoW (not being a king of the hill MMO to rule land) not nearly as much as EVE. Blizzard uses it to combat cheating. [1] http://www.tomshardware.com/news/watermark-screenshots-World-of-warcraft-digimarc-MMORPG,17522.html http://www.tomshardware.com/news/watermark-screenshots-World... [2] https://eric-diehl.com/world-of-warcraft-and-watermarking/ https://eric-diehl.com/world-of-warcraft-and-watermarking/
- michaelcampbell 9y agoCould "fingerprinting" a screenshot be foiled by running the screenshot through a lossy algorithm? Or doing something like converting to jpg->gif->jpg->gif->png->jpg... a few rounds? Enough to keep the picture viewable, but just barely? Or hell, even simpler; taking a cameraphone picture of the screenshot?
- lagadu 9y agoDepends on the type of fingerprinting used. The example I mentioned with the invisible pattern yes but if the fingerprinting is done via other means such as replacing some words on a post with synonyms for each user in such a way that each has a unique combination of words it wouldn't work. Generally speaking in these scenarios you don't want to grab anything directly from the source, just relay what you saw and write it down in your own words. Even then it has to be handled carefully (how many other people got this information? Are the details I'm seeing slightly incorrect in order to filter out who leaked this?) Also take into account that you can be caught just by repeated "A/B testing" of sorts: half the population gets informed that at 19:05 some operation is taking place, the other half that the op is taking place at 19:10. The next day they do the same but using different groups, if you have access to the data that is being leaked you can track down who is leaking it like that after a few iterations.