Nope. Nelson draws his schematics using the Foresthills dialect of the language. In the local argot, a line that crosses straight over another line is a no connection. A line that connects is a dead end, such as the connection of R47 to the +input. Of course, there are incongruities just as in any language, in that C9 and R52 (can't read the number clearly) get a straight line connection instead of an offset one. That's the way it goes. It's a small hurdle that he puts in to test your commitment to building the circuit.
In any event, only the middle MOSFET (IRF610--Q?--number completely illegible, less than optimal scan) is active signal-wise. The upper one (IRF9610--Q?) and the lower one (IRF610--Q?) are current sources, and undriven ones at that, i.e. they are not Aleph current sources or anything of that nature. Nelson is just letting the collector of Q11 and Q12 connect via R46. How much AC goes across there, I couldn't say as I haven't built the circuit, but it might qualify as a sort of backdoor feedback. Note that the connection isn't part of the signal path, per se, as the ZTX parts only serve to set the current and tighten up the two current sources. They are DC servos that correct for AC signals across R44 (Q11) and R55 (Q12). Any deviation from pure DC gets passed back to the Gate of the MOSFET as a corrective signal, thus setting things right again.
Grey