Daily Archives: 2024-03-30

Article note: In everything. The incentive structures around research and research careers virtually guarantee a degree of successful fraud-or-fraud-adjacent behavior, which has been steadily growing to dominate as it tends to be locally incentivized. This study found 19% of medical publications on chronic stress in rats tripped a trivial, long-established standard for likely manipulation looking only at images in the paper. They also found that papers with the features were neither penalized nor localized... and that the garbage level is high enough to change the results of systematic reviews. (Also, I find the method of looking for duplicated/tampered images interesting - magnified insets are pretty standard in image processing research and would show up as suspicious. The benign reasons doing so is common is largely that we still pretend meaningful publications have to be printed on 8.5x11 offset printing, so any image you want to intelligible _has_ to be tampered... once again, let's burn the publishers to the ground and start over on that front.)

Article note: Ever wilder: it now appears the call interception is setting up an RCE for the holder of a specific private key, because login attempts with a specific RSA key would result in the next part of the packet being executed by the sshd process via system(). That's _real_ bad nation-state actor type shit.

Article note: I have a _very_ "all of this has happened before and all of this will happen again" attitude about VCS and especially VCS hosting. Don't get attached, the tools are all awful and the hosts are perfectly situated middlemen to abusively enshittify (like Sourceforge eventually did with bundled crapware).