In online music spaces you can find lots of discussions about quite pragmatic topics like gear, music theory, lots of DAW/sound design tutorials etc. What's always been missing for me is the answer to WHY - for example, what's the motivation behind using certain tools, or using them in a particular way (aside form personal preference)? What do people actually want to say through their music, and how do different techniques help them achieve it? I would love to discuss workflows with other people. For starters, I'll share mine, with a little soundbite that shows my current approach.
Arranging things on a timeline might be easily the most influential feature of a DAW, along with looping clips. It's great at giving an overview of the whole piece and allows smooth blending between parts since clips can start/end at any time. This can, however, lead to uninspiring arrangements and transitions based mostly on mutes and blending parts. I would argue, that the timeline is simply too powerful and encourages some "lazy" behavior e.g. drag that 4x4 kick pattern and stretch it for a couple of minutes. Maybe that's why tracks I make with a pattern sequencer are half the length of those I would make with the timeline sequencer. Moreover, it's very easy to work without intention, by moving things around and hoping for a happy accident. After a while, the timeline becomes like a board in a basement of some conspiracy theorist - an intelligible mish-mash of audio, MIDI, and automation.
Opposite to this, pattern-based sequencing forces you to be more intentional with the changes, because they're harder to make. It's harder to see things, and the notion of time disappears completely (is it better to not know how long the piece is?). At least one won't feel lost and overwhelmed, since things are kept in neat little packages. To some, it might feel more organized. It took me a while to find the most optimal way to use patterns and tracks with different lengths and step scales - it's something that wasn't so obvious with the timeline. With Elektron sequencers, one pattern can form a whole part of a song: setting note scale to e.g. a half-note lets you create full progressions. Also, managing automation is easier, since it belongs to the pattern unlike in a DAW, which keeps it global which makes it annoying when moving clips around.
There are two completely different approaches to organizing tracks within projects. In the DAW world (and in some hardware sequencers), the same tracks are kept throughout the song, so if you want to have a different preset on the same synth in verse than in chorus, you might have to create a different track for it. Others keep unique track settings on a per-pattern basis, like electribe or model:samples. At first, I found this idea horrifying: how do I manage changes I want to make to a track? Do you need to copy-paste it to every pattern? I ditched DAWless because of it, only to come back to it because of this exact feature years later. I had to let go and enjoy the ride, stop obsessing about controlling everything. Now I see it as a great advantage, an invitation to make each part of the song different and surprising.
This brings me to the answer to "What are tracks for?": a track doesn't have to mean one sound, not even one instrument! To take it further, how do you even define an instrument? When I started making music, I followed a very strict blueprint, expected all songs to go more-or-less like this: drums, bass, synth pad, synth lead, maybe some percussion or vocals. It's a concept derived from old rock/pop music, where musicians don't change the sounds drastically during the performance. This has obvious limitations and frankly, I can't listen to some of those records anymore.
In comparison, in the world of samplers, anything can be an instrument, any sound is valid. Since it's easy to use a much larger variety of sounds (even on entry-level devices) you can always surprise the listener. There's no need to recreate 1980s setups that had only 1 type of kick drum or synth pad for a whole 5 minutes. You can think of it as a shift towards non-representational art - previously known roles such as "bass", "pad" or "lead" can be interpreted very loosely, or ignored at all.
On a sampler with a limited number of tracks (like 6 on model:samples) you can see how much it will improve the visibility of instrument interaction: placing kick drum and bass on one track means that they won't clash - they simply can't play at the same time. It will be easier to change the bass according to the kick without jumping between tracks and trying to remember which one plays on which step.
Just because you can, doesn't mean you should! It seems obvious that garbage in = garbage out, especially when you're recording a band in a studio. But what about synths, samplers, and plugins? They can produce much cleaner signals than microphones and amps, so it might be tempting to assume that whatever gets recorded can be mixed easily.
That's just the tip of the iceberg though - what I failed to see is that mixing starts before anything gets recorded. I would argue that the pitch is the most important mixing tool available. If the instruments are not fighting with each other in the same frequency range, then there's no need for so much work with the EQ in the post. In the case of one-shot samples, tuning is a great option to create distance (frequency-spectrum-wise) between the sounds. When that's not enough, the built-in filter (I assume every sampler has one per track) can do the rest.
Another layer I've started experimenting with is the timing of notes. In the realm of electronic music, it's very easy to have two sounds occurring at the same time (since you might be working on a quantized grid). That can be a good thing, if the sounds are complementary, e.g. a soft low thud and a higher-pitched transient. However, if they don't gel (there's no synergy) maybe it means one of them should just be... removed. Think about linear drumming which is "monophonic", i.e. only one sound occurs at a given point in time. Quite often the adage about removing until things can't be removed anymore rings true. If I don't want to remove that note though, I try to move it e.g. with micro timing (like 1/96th) to see how the relationship with the other one will change.
In the post-ultramaximizer era we have to repeat ad-nauseam: don't kill the dynamic range! Think of volume (MIDI: velocity) as another means of expression. It has to be intentional. How can you surprise the listener and create tension, when everything is at the same level at all times? The roles can change dynamically. Who's the lead, and who's the supporting cast? Perhaps with careful planning at this stage (and others) you could just record the stereo output of the groovebox straight to the computer and ignore mixing (not mastering) completely... so, to summarize: your composition is already your mix!
Working with a limited system has also a great advantage of discouraging endless tinkering. Over the years I've come to realize that if a song sucks it's probably not because of a hi-hat that's 2 dB too loud. No matter how many effects I put on that guitar, if the part is boring or played badly it will still suck. I feel that because of the nature of interaction with computers, it's easy to waste so much energy on details and completely lose track of time, instead of focusing on the source material. It's all about those bold brushstrokes. It takes a lot of courage to delete whole parts that don't work, try totally different approaches etc. but it might be worth it (in contrast to fighting a lost cause). Depending on your personality, you, like I, might need to commit more to finishing things, even if they're not perfect. I welcome exhaustion with open arms - it's a sign the track is done and it's time to move on. Enough is enough.
After many years of collecting gear, I have this sad realization: probably I could have made amazing things with the basic setup I had back in 2011: a tiny MIDI controller and a lite version of Ableton 8. But I didn't. Why? Well, it does require a lot of imagination and focus on available tools/features. Instead, I was moping about not having the powerful things other people had. My biggest pain point was the lack of good-sounding instruments and effects - anything heavier than Synth1 would seriously slow down my laptop. All I wanted was to turn that filter knob and hear the change in real-time. I didn't see the potential a single sample track can have when sequenced with a lot of automation. Even the oldschool Synth1 can shine bright if sequenced properly! Similarly, Ableton's Simpler, Drum rack, or Impulse can offer quite a bit of variety and more than 1 sound type per track, so the limitation of only 8 tracks available (lite version) is not such a big deal.
Ironically, to understand all of that, I had to get away from the computer. Perhaps the nature of that kind of interaction inhibits creativity in the early stages? With a real instrument, you're literally touching music (especially chordophones). You can become one with it. There are no endless presets to scroll through. The reaction is immediate. There are no distractions, no web browsers with chat or whatever working in the background. On my grooveboxes the options were always limited, so one has to learn how to squeeze the maximum out of it. They can be also quite opinionated in terms of workflow: MPC, SP-404, and Digitakt are completely different and I dare say, they will nudge you in different directions. With a DAW it's a little bit different - I can use Ableton to record a rock band, make generative ambient or lo-fi deep house. Since Ableton or Bitwig offer meta-devices (containers) and lots of other complex features you can build many different workflows. But how can you create one, if you have no frame of reference? For this reason, I would argue that being exposed to a more limited environment works wonders. Notably, Elektron machines offer a great learning experience with their parameter locks and sample locks. The same can be said about Novation Circuit with its very hands-on sequencer but a completely different approach to track/pattern management than Elektron's. Being exposed to different paradigms is a great thing - seeing how others solve the very same problem can help you develop your own solution, or take an existing one to your DAW.
This is the most important principle. For years I've struggled to recreate music that I've liked so much, and failed miserably, only to realize, that I was barking at the wrong tree. Forget the genres. Time to create a new one! Who said you can't mix reggae with black metal... Music theory is a useful thing to know, but the more I stick to scales and try chord progressions, the more obvious the songs are, to the point of being silly or cheesy. I also refrain from using any particular structure like verse-chorus, it's just too strict. The point is to have fun, experiment, and express yourself. Therefore, anything goes.