Yes of course the ratio is constant. However, it is only a matter of time before mining with any hardware simply becomes a waste of electricity.  I wonder if it would be possible to "force" the difficulty to decrease by coordinating a pause in mining, then quickly mining the reduced difficulty. Then the process would be repeated.

As the Bitcoin network difficulty continues to skyrocket, I find it difficult to believe that mining will be effective. Even at this point, I wouldn't even attempt mining without at least a couple hundred GH/s. Even then, I might never see a ROI. It seems to me that by the end of the year (probably much earlier), any hashrate under 750 GH/s - 1 TH/s will probably fail to produce a viable source of bitcoins. This solution could be solved if hardware prices were eventually slashed by as much as 40-50%. Only a fool will purchase hardware that has no hope of producing a ROI. Either the hardware hashrates increase dramatically (keeping the same prices), or the prices must eventually be lowered by a large margin.

If a 1 TH/s miner is purchased for around $2400 (rough estimation), it would still take ~134 days to break even, assuming a power consumption of around 2000 watts. By the time this hypothetical miner actually pays for itself (assuming it doesn't suffer heat failure or another issue, requiring a warranty replacement and wasting valuable time), it will be comparable to the BFL Jalapeños of today.

Please correct me if I'm wrong, but it doesn't seem like mining will be a viable investment by the time 2015 rolls around.