I think it makes more sense to have chambers have their strength based on the size of the chamber itself, but you need a certain number of crew to keep a chamber running at maximum power. Otherwise it starts to lose effectiveness, and maybe even starts suffering random outages.
As far as we've all looked at this so far, I'm in agreement with the above. It's more nuanced and complex, but also intuitive. It allows systems to complement & complete each other. It's nice because none of this affects core ship functionality, it only affects optimization. A new player could still slap together a basic ship with simplistic chambers and have a fully functional ship with decent stats, but it wouldn't be able to stand toe-to-toe with a dedicated PvP ship that had a well-organized interior and robust crew. The level of complexity from a system like this would mean that to get anything above the grade of maybe a frigate or basic destroyer up to
fully optimal statistics would probably require a team effort by multiple builders over a substantial period of time, which would raise the value of optimized capital ships to epic proportions (probably to the point of warranting self-destruct systems to avoid losing the strategic edge). At the same time, even a fairly new player would be able to engineer very well-optimized fighters, bombers, scouts, freighters & corvettes.
To follow up on a few grey areas that seem to be raising questions, I'm going to talk a little more about my thinking on this after looking at the excellent ideas and that other thread from earlier.
My initial concept of this was definitely as Lecic describes; that the chamber strength derives primarily from the bounded size, almost exactly as they currently function except becoming hollow. Much of the chamber would become invulnerable empty space in essence, but damaging the bounding arm/frame would probably do crippling amounts of damage to the chamber, rather than just a percentage damage equal to what those blocks represent of the empty space that defines the chamber.
So most of the chamber's function would be wrapped up in the amount of space bounded. From there, I imagined that its effect would be boosted further by decor (e.g. perhaps a 100% value increase to the chamber for optimal decor numbers). After that, if crew is implemented - in any form; NPCs, sprites, or even GUI-only - essentially half or more of the additive value of decor would become related to crew.
EXAMPLE: The light reconnaissance frigate 'Razvedchik' has strong stealth, detection and speed capabilities. It's natural FTL jump charge speed is 120 seconds (e.g.). The designer runs a hall from engineering and frames a new FTL chamber nearby. It reduces the charge speed by 25% to 90 seconds. Another designer then installs interior hull, workstations, decorative walls and panels, flashy lights, etc, doubling the chamber's value to 50% and reducing the charge time to 60 seconds.
While scouting an enemy mining operation, The Razvedchik is detected and takes heavy fire from drone escorts while attempting to jump clear. One shot passes through the FTL chamber, damaging some computers and hull, but not touching any of the actual FTL chamber bounding blocks. Charge speed increases 3% (based directly on the damage % to the chamber decor only). Just before getting away, a rocket breaches the outer armor plates and vaporizes 10 blocks of the actual FTL chamber bounding arm blocks as well as more decor. The decor damage reduces chamber function another 7%, but since the bounding arm only consists of 40 blocks to begin with, the chamber loses 25% of its base function as well as perhaps a temporary (60 or 120 second) complete outage while systems are stabilized, setting the jump drive recharge back to 2 minutes for now, then around 80 seconds once the FTL chamber is stabilized.
Something along those lines anyway.
Obviously, there are many ways such a system could work, and probably better ways than what I'm thinking of right now. In fact, I'd prefer to see some kind of hull-integrity modifier, but to look at that for an initial systems seems excessive. I'm looking mostly at what an initial trial implementation could look like. I like all the good long-term ideas coming out though!