Vostok flight computer had a number of preset programs that could be activated by receiving the appropriate command from Earth. The COMSAT receiver was based on an older military system that was originally designed to remotely detonate explosives. Soviets widely used radio-controlled mines (called F-10) in WW2 to destroy strategic targets left after troops withdrawal. That system, designed in late-1930ies listened for a few preselected radio frequencies and waited for a specific sequence of impulses to arrive. It was not particularly well protected against interference and there's even anecdotal evidence that Finnish troops used a song played over radio in an attempt to jam or detonate these mines. After the war, the system was adopted to control military ballistic missiles and eventually made its way into Vostok spacecraft with its roots in ICBM program. So a radio system used to select which particular mine to explode became a radio system used to select which particular program to run. Of course, it was modernized along the way, with one of the notable improvements being a larger number of possible commands. But still, every command was nothing more than a set of short radio impulses on specific frequencies. If a signal was received on specific frequency, it was a logical 1, if no signal was detected on that frequency that would be logical 0.
Then, in 1965, a would-be-Vostok-2 was running several tests to test a brand-new and never flown before equipment for EVA. That was an unmanned mission to avoid putting cosmonauts at risk because EVA equipment was potentially very dangerous. Mission control would slowly run EVA operations step-by-step and record state of a spacecraft after each step. Unfortunately, they had only a narrow window of opportunity to communicate with spacecraft when it was flying over Soviet territory. Soviet radio equipment was not particularly reliable, so they had a primary station and backup one in case the primary one would fail. A range of those stations was also relatively short, so different stations would be used on different orbits (and some orbits would have no SATCOM at all).
Now let's move to the Kamchatka peninsula that happened to be close to the path of next satellite orbit. Mission control was well into the test program, with EVA dock already deployed on spacecraft and it wanted to remotely open the hatch to EVA dock. To do so they had to transmit command #42. That command originated in main mission control center, but had to be relayed to spacecraft by Kamchatka radio station NIP-6, with another station NIP-7 serving as backup. It was expected that backup station would stay silent, but due to management mistake, both primary and backup stations were active and therefore relayed command #42. But relaying a signal introduces some (small) delay (for example there's a delay of signal propagation over distance), so spacecraft received two slightly shifted copies of command #42.
Now consider that a command #42 was a sequence of impulses that we can write, say 101001. Remember that lack of an impulse is "0" and presence of impulse is "1"? Impulses of second, delayed copy of command #42 were interpreted as "ones" in positions that were originally supposed to be "0". Like this:
101001
+ 101001
= 1111011
So instead of a command 101001 (decoded as "42") spacecraft thought that it was command 1111011 (decoded as "5"). Note that I took those particular bit sequences out of the thin air for illustrative purposes only.
Unfortunately, command #5 happened to be "start spacecraft descend". Executing this program would take spacecraft back to Earth, but since it was positioned on orbit in the wrong position, it would not descend to USSR territory. Soviet designers anticipated the possibility of such scenario (descend of a secret spacecraft in enemy territory) and installed independent self-destruction device that would destroy spacecraft if it attempts to descend outside of the pre-programmed range of valid descend trajectories. Therefore soon after command #42 was send by mission control, communication with satellite suddenly and abruptly ended. Recorded logs later showed that spacecraft relayed confirmation of "command #5 received" and the whole story was eventually reconstructed.
In aftermath of that event later Soviet SATCOMs were updated. However 12.12.1975 another satellite Kosmos-785 was lost when a random error in a normal code sequence was misinterpreted as a self-destruction command.
More on that topic (in Russian) can be found here and here.
