I was looking for the independent testing of FSRS 5. Sadly testing alpha stage applies for only one preset at a time, so making it apply to all deck presets is time consuming and not very efficient.
However I did not notice that short-term (<1d) scheduling in hours is not enabled, as I thought it would do. I thought this was added to FSRS 5 
Short-term (<1d intervals) scheduling with FSRS-5 isn’t avaiable yet: Let FSRS control short term schedule by L-M-Sherlock · Pull Request #3375 · ankitects/anki · GitHub
It’s a work in progress, and it’s unclear when or if it will be implemented.
Ohhh… that explains it
Well, it wasn’t rejected by Dae.
Right now, the PR for disabling learning steps by leaving the field empty has already been merged, so don’t worry about that. But the PR for letting FSRS schedule <1d intervals is still a work in progress. It’s not clear when it will be done, and it’s not clear whether Dae will want to release the next version of Anki before this is implemented or after.
That is nice to hear…I don’t know how this all pans out for all the difficulties that might ensue. What is the chance of being able to set the again interval manually, but allow the rest of the (re)learning steps to be scheduled automatically (provided that idea is indeed feasible and it goes all the way to the end)
You can’t do that. You can experiment yourself - try different learning steps and see that they affect both Hard and Again (in the learning stage).
But you will be able to enable learning steps while disabling re-learning steps. Or the other way around. Or disable both. Or neither. Whichever combination you like.
Hmm yeah, I see…What I was hoping for is locking the interval at a custom interval, say 1s, always so that pressing again always makes the card show after 1s. Upon pressing Good, FSRS would then take it from here. Kind of like a hybrid method. But I guess that does more bad to FSRS than it does good…particularly since it is in a prototype phase at the moment.
If you only use Again and Good, then just make 1s your learning step. FSRS will control the first interval for Good
Really??? That was what I was hoping for :DDDDDDD
But you won’t get a subday interval. Is that fine with you?
Oh. What I was hoping for is the FSRS is for then FSRS to control it all the way up from Good, <1d or >1d, like how Expertium put it. I dont use Hard at all. I mostly use the binary of Again and Good 
There were two PRs:
no learning step = 1 minute learning step when FSRS is turned on. This got merged. Now, “no learning step” behaves similarly to how “no relearning step” always used to behave.Let FSRS control subday scheduling. This is not merged yet. So, if you have no learning step FSRS can’t assign it a <1d interval.
Yes I am aware that at this current point of time FSRS 5 subday scheduling is not even done yet and may take longer than planned. So technically speaking, FSRS wont schedule any subday interval at all. It would just remain idle.
I was just talking about what would happen if FSRS ends up being able to schedule subday intervals, despite having a fixed Again interval specified, meaning that FSRS controls Good intervals while leaving the Again intervals fixed. Expertiums concern was that Hard intervals also depend on Again intervals but I don’t use Hard buttons at all : P
It will get you a 1d interval instead.
Currently, Hard interval is equal to first step if that’s the only step configured. Not sure how exactly subday interval will be done, but probably after the first step, cards will still stay in learn state, so hard should still show the same ivl? Idk.
FSRS will change intervals after every review, including same-day reviews. So if you have a 1s learning step, it will affect the next interval given by FSRS.
In what way? Would it be for the worse or better, or just different…? If I could maintain a fixed Again interval, where FSRS is being allowed to do whatever it wants with Good intervals at the same time, I can do with a little bit of compromise in case of having a fixed Again interval is suboptimal.
If you clicked Again, then the next interval will be shorter.
Btw, I don’t think it will be possible to have <1d intervals AND learning steps. I’m pretty sure that if you have any learning steps at all, FSRS will not schedule <1d intervals.
So if the optimal first interval for Good is <1d, but you have any learning steps at all, the first interval for Good will be 1 day.
Oh well…
If you clicked Again, then the next interval will be shorter.
Does this mean that it will generally approach 1s if I keep on pressing Again multiple times.
Yes and no. It depends on your ratings yes, but I don’t think you’ll ever go down to 1m or 1s. Humans have a STM of 30s so you of course won’t forget cards in 1s.