Would you join a new storage-based mining pool, and if so, to what extent?

There are many different ways to mine bitcoins (CPU/GPU/FPGA/ASIC), but nobody seems to have written a Rainbow-table solution yet.

I've worked in Crypto for decades, and I thought I'd take a crack at building such a beast, but I need to know how big I can sensibly make my tables. This is all currently theoretical, but please can you take my poll and let me know how interested you would be.

New 3tb hard drives are about $136usd today. You can hang upto 100 drives off each USB host controller in your PC (if you've got enough powered hubs and cables etc). Usage after initial setup/pre-computation is very low bandwidth.

Here is a rough cost table, showing terabytes you could make available, the $USD cost of that much in hard drives, and the monthly power costs to run them (assuming 15c/kwh - ignoring all controller/hub/PC costs)

TB   Drive $   power $/mo
3     $135    $2
6     $270    $3
12    $540    $7
24    $1,080   $13
48    $2,160   $26
100   $4,500   $55
200   $9,000   $110
500   $22,500   $274
1000   $45,000   $548
2000   $90,000   $1,095
5000   $225,000   $2,738

Assuming that any investment you make will produce bitcoin rewards that are the same as, or better than, current ASIC predictions (eg: performance better than ~ 1.5Mhash/Sec per ~ $150), how much storage (if any) would you be prepared to invest in such a pool???

Note: you will need a permanently-on PC and permanent internet connection to do storage-based-mining - the actual mining "Brain" would be needing to infrequently query your storage for index lookup results. (and, if you're running a "Brain" on your PC, it will need to query other pool worker storage systems for index lookups itself).

Storage-based mining using Rainbow-Tables would be a new and different way to mine, and involves enormous quantities of storage (disk drives) holding pre-computed lookups.

In laymans terms - a rainbow table is a bit like an index in an encyclopedia. Current bitcoin mining works using MegaHashes-second, which is kindof like how fast it can search through an entire set of encyclopedia volumes, looking for a specific thing. Rainbow tables work a different way - instead of searching fast for something, they compute a pre-loaded index of candidate starting points - or basically - they are used to "jump" directly to the approximate location of the answer you're looking for, instead of having to search mindlessly for it.

Once the tables are pre-computed (an offline setup process that takes several days), drive use and access is very low bandwidth (the "Brain" controller software seeks only one specific sector of data out of all it's available space, infrequently.

Thanks for voting!