that’s incredible 
thanks for alerting me to this!
To me there’s nothing more natural in music than a basic oscillating wave. Synths are the most natural sounding instruments because its pure energy and math.
You have total control of the timbre and amplification envelope. Shaping those aspects of the sound is where the “emotion” comes from
Just my few cents … and only IMO … 
I suggest to split the challange in parts and consider topics like:
- human touch in general
- supporting hardware
- samples vs synthesizers
My thoughts aren’t scientific or comprehensive, but maybe they inspire to get an idea.
Human Touch in general (absolutely not comprehensive):
- no perfection
- small fluctuations in timing, pitch, accentuation, and sound, which often are so tiny that we only realize them if “something” is missing … like human vs machine, analogue vs unnaturally clean
From my experience it is hard to simulate this by just using a randomization function, a clean LFO shape, or the typical MPC swing - which to be honest, comes close in some cases.
It maybe too simple to say, but if we understand the way an instrument is played and sounds, and how a musician would interact with the instrument, we can pretty nicely “simulate” human touch and emotion with many synths or samplers.
This said, either I play parts live on a keyboard or I use a sequencer, which supports editing or recording of tiny time-shiftings and sound-fluctuations.
As an example on an Electron, a MPC, an OXI-One etc we can record movements of parameters in real time. Recording parameter changes with personal rhythm and expression along musical phrases creates much emotion even if all notes are straight on the grid. Now add some timing-fluctuations carefully … for me this works
Supporting Hardware:
- a real instrument interface like a keyboard, Linnstrument, a multi parameter controller - Roli Seaboard or Osmose Expressive comes to my mind - and there are many more
- for human touch I would prefer a MPC like interface with softpads, which I play live over a typical groove box with 16 on/off buttons to “program” the pattern.
IMO to play an instrument live will more or less automatically help to generate expression directly without planning, programming or editing. If someone is capable to play a keyboard this can be a shortcut.
Just to have it mentioned. If sequencers shall be used for achieving human touch my choice would be those providing a “linear sequencer” (MPC) rather than a “step sequencer” (Elektron).
Samples vs. Synthesizers:
If it sounds good - use what you like, be it single-shot or longer phrases for samples or a synth. For vocals I have a selection of samples providing single vowels, phrases and improvisations. I chop them, put them through FX and achieve at least the perception of human voices singing …
My suggestion would be using “modulation” (see “not perfect” and “fluctuations” above). Modern synths have many options for modulations. Used like spices in tiny amounts can make the difference between sounding clean and machine-like vs natural and emotional.
IMO use of natural preceived voices gains a natural or even human touch. The source is not important, it’s the natural appearance - if this makes sense to you. For this I make use of some Eurorack modules, which can generate this. My favourite is the XAOS Devices Sofia. This module can’t talk, but it’s built from analogue circuits, which have been designed to create formant like sounds.
I hope this wasn’t too long to read and not too theoretical …
Not trying to start a debate, and this is a little off topic, but some people in the psychology world have suggested that loop based music is really soothing to people with PTSD and similar trauma based issues precisely because our lizard brain finds it so predictable, which seems pretty dang human to me 
LFOs, P-locking different values per Step, Recording Live parts.
I’m quite sure the examples you mentioned contain ingredients that apply to my theory of not feeling “human” but exactly the opposite, the “robot” like groove which is a “imperfection” and therefore can evoke plenty of emotions.
Can’t guarantee a complete lack of samples in these but here’s a few that spring to mind when I think of human (electronic) music- I guess something in. the elasticity of the synthesis and/or sequencing, or the playfulness of how certain elements move around each other, idk
Lol. So according to you it needs to be imperfect to be ‘human’ but if it’s perfect that’s an imperfection and is also ‘human’. Nice try.
Humans made these machines. Humans play these machines. Other humans listen to it. It’s all just different expressions of humanity.
That’s all.
Not quantizing, using tempo changes and tempo swells, using the pitch bend are some ways.
This can make it purposeful, and also ensure that you don’t suffer from option paralysis. It’s fun to make foley sounds. You can record them with your phone, and load them onto a cheap sample player.
Yes! Came here to say check out Terry Riley if you want electronic music with heart. I’d also recommend Tod Dockstader who was really a total outsider in the electronic/musique concrete world.
Also, I’ve noticed that a lot of the really great electronic producers (Drexciya, Larry Heard, Aphex Twin come to mind) do a lot of live playing over sequenced elements. Adds a ton of life to their tracks.
Microtiming.
Play drunk, mix sober.
I kid, I do find that, say, poly AT or MPE can add a little humanization to a performance just because of the physicality of either method of playing. I definitely don’t think of MPE as a mechanical or repetitive technique.
may i introduce you to the post-rock genre
if digital is less than human i guess the only solution to get a feeling is to take digital and bang it up a bit yeah?
regarding samples, I used to feel the exact same way, I can’t in good conscience take sounds i didn’t make and just “alter and arrange” them and really call it mine. until i realized the whole purpose of making music is to share a feeling, or perspective. taking a recording and altering it, and arranging them in time and spectrum, is literally what your eyes and ears do to develop your own perspective. i can’t very well convey much perspective with oscillators. I guess you can, given dedicated study of intervals and composition, but you have to really know what you’re doing. sampling opens the door up to convey musical perspective where anybody can do it, knowing nothing. that’s why the advent of the MPC is such a big deal, it grants access for free. if it’s not satisfying, that’s cause what you really want is a challenge, that’s where you dig into learning how to do it the hard way.
just this 
I’ll challenge that.
Some drummers can ‘splash’ the hihats using the foot. You get a closed and open sound at the same time.
I learned it by accident.
Not the same thing as hitting with a stick, closed and open, but it can be done.
have been in love with this album since I picked up the reissue a few years ago
…by adding all sorts of dirt and imperfection…
what makes a massive attack track so close to home…?
the dirt track…pretty much all their big ones, had some noise/tapehissing/natural athmosphere foley thing going on under at least one fader, that was driven in any way but static…
dynamic volume automations on all sorts of elements, a few here and there not dead straight quantised elements and a little dirt underneath will do ALL the trick…promise.