I just tried this and messing with the little probe seems to have caused it to work once, then it went back to not working after that.
I think in my case the problem must have been with the card rather than the camera. (Unfortunately I don't have another camera to test this in and am not sure how to detect whether the card is locked on my desktop because it seems to be able to write to the card whether it's locked or not.)
I compared the problem card with another card and noticed that the notch below the lock "switch" seemed slightly shallower than the working card, so I switched the switch to "lock", took a pocket knife, then scraped just a little bit of plastic from the area that the switch occupies when it's "unlocked". This enlarged the notch just slightly (maybe 0.1mm? It wasn't even 0.5mm) and now it seems to be detected as locked.
Oh, and if anyone decides to mess with the probe thing, make sure to take the batteries out. It may look like the camera is off when it's off, but it's not really off. "Asleep" might be more accurate. The power buttons generate an interrupt that the cpu has to handle in order to turn the camera "on", and so the cpu has to be "on enough" to be able to do that. Who knows what else is powered up. I failed to take the batteries out of mine and it was acting weird after I powered it up. I had to remove the batteries for a few minutes to get it back to normal.
FWIW, I took a twist tie, stripped a little plastic off of it, bent the wire at a 90 degree angle, trimmed it down, then clamped it in an x-acto knife handle (in place of the blade) to make it easier to hold and manipulate. I then used a tiny flashlight to illuminate the slot. Not as difficult as it sounds, and the x-acto knife probably made things much easier.