Sorry if it sounded like I was arguing with that. By correctly riding the beta and saturation curve you could get it to oscillate with a resistive load, but it's still not a great circuit....
But this dos not change the fact that this circuit will work even if you replace the speaker with a resistor. If you choose resistor wisely this circuit will work without the speaker.
...
What if PSU voltage changes and causes base current to increase? Or if someone changes the resistance of the load, or replaces the transistor with a different transistor?
Both of those circuits you just posted are proper multivibrators where each transistor turns the other transistor off. They are much better and should run well, relatively independent of PSU voltage or transistor beta.
If you added a resistor or two you could add hysteresis to the Q2 base turnon/turnoff point and make it a really reliable circuit.