I still say the "technically correct" thing to do is for every potential observer to make 3 perception tests...
I am not getting upset if people rule that each individual guard get to roll perception multiple times against each intruder and that each intruder reroll their stealth against every single guard they run into (and in some cases perhaps even multiple times against each guard).
I just don't think this was what the authors had in mind when they deliberately changed sneaking, invisibility and silence to set 3 different thresholds for perception before the infiltration was even started.
The astral perception table have list 17 items over 5+ thresholds, yet I am pretty sure you just take one astral perception test to resolve that.
From my understanding, your approach is to make failed Stealth test less impactful by making the guard suspicious instead of raising the alarm.
Ah, now I think I understand what you mean! Sorry :-)
No, my approach was that:
A guard that See an intruder (for whatever reason and no matter if he also Hear the intruder or not) will raise alarm.
A guard that Hear (for whatever reason) but does not See (for whatever reason) an intruder will become suspicious.
A guard that doesn't See (for whatever reason) nor Hear (for whatever reason) an intruder will not react at all.
Reason why the guard 'Heard but not See' the intruder in the example was because invisibility was successful but stealth was not. But the guard would probably also be suspicious if the intruder was walking around like normal in the next room, or if the intruder was walking as normal behind a guard because he had a silence spell but the guard resisted the silence spell, or if the intruder was walking in front of a guard while using invis and silence but the guard resisted the silence spell. Etc. There could be any number of reasons. Not all of them depending on failed stealth specifically. Failing 'Stealth' was not really the key here. Guard 'Hearing but not Seeing' the intruder was.
Stainless Steel Devil Rat means to give actual Edge points, which is true to the rules. However, it doesn't fit perfectly well with the idea of a single Stealth test that fits all circumstances in advance.
With few exceptions, outside of combat and hacking Edge from *gear or qualities* etc typically either just award one point of Edge for the entire encounter / scene or they reward you with one bonus Edge point for a specific test that also need to be spend directly on the related test.
Since book is pretty explicit that you typically just roll once at the start I think the intention here is actually that you do just roll stealth once (just how you typically just roll spellcasting once for invisibility or silence = streamlined) and reward one edge if you have coating. This also opens up the possibility to have a larger patrol of guards to roll perception once as a Grunt Group (rather than two-three opposed rolls for each member of the patrol). This me and SSDR *disagree* on. Which is fine. There is no fixed rule either way I think. But still.... automatically earning 2 Edge every time a pair of guards walk pass you...? That will pile up a lot of Edge very quickly(!)
Anyway.
But then I also *agree* with SSDR that both parties may gain additional edge during the infiltration attempt based *on circumstances*.
Trying to sneak pass a checkpoint that is well lit and there is little cover will be harder than normal and will probably grant a point of Edge to the guard at the checkpoint. Sneaking pass a guard in a large warehouse with AGVs moving all over the place, is poorly lit and have a lot of cover will probably instead grant a point of Edge to the intruder. No issue here.
What to spend Edge on then? Well.... Guard can spend Edge on his perception test. Intruder can also spend Edge on the guard's perception test. I don't really see the issue with this to be honest.