... in practice it makes many skill checks almost GM fiat - I have had to write down some of my calculations to try for improved consistency, which ends up eventually losing the freeform ideal.
Sorry I missed your post. Supplement Four responded pretty much like I would.
Fiat is, IMO, doing something without reasoning it out - consistency is also fiat (ex: always getting +1 per skill level is fiat since it does not account situationally for anything). As a Ref I am tailoring checks to match a given situation, the limits of 2D6 and even a particular character as developed by the Player - even from prior events in the game.
Every check may be unique. For instance, if you are running from combat down a long hall and the door slams open at the other end disgorging a running armed assailant... if you try to shoot him with a laser pistol (not RAW, but standard IMTU) then you are probably gonna hit him - he's silhouetted in a confined space and all. With the laser, good chance you will neutralize his weapon threat if you can see him, despite running. I'm not gonna run that as combat - its one or two rolls, depending on length of hall and the aspects of the NPC, for effect (hitting being a given unless the rolls are a complete miss). If the PC has experience with the weapon he's got a good chance of success, regardless of range. Unless he is wounded, acrid smoke is filling the hall, he's just escaped an explosion, his party is pushing him, etc. etc. Lots of variables, all accommodated by a simple target selection and usually just one DM.
The problem with a DM like this is setting the difficulty so that a noob has a chance of success, and someone with skill-1 has a good chance of success, but someone with skill-2 or 3 doesn't just automatically succeed. Do you use some form of autofail? Even then, some consideration for skill needs to be made, because a failure by a noob or even a skill-1 might be spinning off into space, but someone with skill-3 is not going to do that even on a failure - at least not 1 out of 36 tries, which is what most autofails would be.
No autofails for me generally (though limited success is another story) - but auto-success sure. Unless there is a good reason for a Vacc-1+ to make a check, they succeed on basic activities.
Using the RAW, non-ordinary 'dangerous' activities (things the PC is attempting to do - not things happening from the environment) are automatic successes for Vacc Suit-2 (+8DM with 10+ target). If say, however, a missile impacts nearby, causing depressurizing of a nearby hull and flying shrapnel - they will tend to get something more along the lines of the RAW 'recovery check' - which is +2 per skill level with a 7+ target. (Depends on how close things are - which I would tend to determine by a throw - not fiat, though the target is chosen by me...)