Incidentally, the parallel diode, if matched, would give a circuit something like a current mirror, but since it's floating, it would act more like a transistor with really low (but fairly stable) hFE, equal to the ratio of areas; when done with matching transistors (one diode being a "diode strapped" transistor, B+C tied), hFE = 1.
This obviously doesn't help any, and yeah it seems likely it's supposed to be backwards.
There's also no bias current through the transistor (or diode), unless there's a negative supply elsewhere, which seems unlikely given the trouble of a whole transformer and rectifier here, with no supply symbol distributing that voltage elsewhere.
The diode would at least constrain the output voltage to within +/-0.6V of the setting, ugly but better than nothing I guess.
And the collector load RC is... irrelevant? Maybe it extends SOA by soaking up some of the voltage drop; it also serves as a current limiter, but a slow acting one given the capacitor.
For grid bias, typically a fixed voltage at high impedance (10k+) is adequate, and this suffices for class 1 amplifiers; class 2, you have to source some grid current, which might be the intent here, but it's really sketchy doing it without any idle current through the transistor at all.
Tim