Having read and studied many messages lately about various modes and
advantages of one over another. BPSK, PSK31, SLOW CW, NORMAL CW etc
My conclusion like I have always stated is that NORMAL CW or SLOW CW is
more suitable on LF for what we are trying to achieve, especially long
distance dx, and short exchanges of information like callsigna and RST.
If one wants to shift large volumes of traffic then use data modes and
suitable frequencies like vhf/uhf and satellites, where noise is mimimal
and phase distortion due to propagation anomalies is also mimimal.
Propagation conditions on LF with all the environmental noise and high
static levels that most users of the band complain about plus low
transmitter power of a max of 1w is more suited to the use of
CW where a mimimum of fuss and equipment is required. The operators
ideally should be fully competent and experienced in CW.

On the question of machine v manual modes.
If one uses a computer to visually read slow cw that cannot be heard
aurally then it must be a machine function system, otherwise the message
or qso could not be conducted.
In most cases the very slow cw that I have seen on a screen can be read
directly from the speaker and in this case is not considered a machine
system. If one is totally dependent on a machine to send and receive
very slow cw then it must be considered a machine system.
The machine/computer with suitable software is converting the audio into
a visual format, and with FFT,  processing the signal. If automated
sending is used, again this is a machine generated system.
The operator participation using a computer for sending/receiving very
slow cw is no different to sending/receiving PSK31.
You must read the screen in both cases to get the message and you must
type the message information to send it.
Another factor to consider. Very slow morse generated/received by a
computer does not require the operator to know the morse code. The
message is typed to be sent and instead of reading the received dots and
dashes on the screen, existing software can be modified to print out the
message.
A couple of professional cw operators one at each end of the proposed
transatlantic link will achieve the object. There is no need for
synchronised atomic clocks and bandwidths of 0.00000000007hz
Keep it simple and all can enjoy and verify it.
If someone using the new Hydrogen/Stentlesch clock and a b/w of
0.0000000000000000000007 hz told us that they had made the first qso
across the atlantic on 136 khz, how would the rest of us  know if they
were telling the truth. Would they qualify for any awards?????
73 de G3KEV