Here's an analogy: Suppose you take an old car with rear wheel drive and chop it in half such that the front part, that contains the motor, is completely separate from the rear part, which contains the differential and rear wheels. Now, connect the two together using some steel rails and such then turn the key. The motor revs up but the car ain't going no where.
When the car was intact [before cleaving it in twain], the front power source was connected to the rear drive by a drive shaft. The junctions in a bipolar transistor are like that drive shaft, in that they join the beast into a functioning unit -- functionality requiring that connectivity.
Or think of it this way: A bipolar transistor is either PNP or NPN.
Back-to-back diodes are: PN-NP or NP-PN, or more precisely, PN-conductor-NP, etc..
Not the same thing ;)
Kind of like:
Whole car: Motor-Driveshaft-Rear Wheels
Car split in half and joined by steel rails and such: Motor-1/2 Driveshaft-Steel Rails-1/2Driveshaft-RearWheels.
With a PN-Conductor-NP arrangement, if you hook it up like a transistor, you'll get some current through the PN diode but no amplification. With the Motor-1/2 Driveshaft-Steel Rails-1/2Driveshaft-RearWheels arrangement, if you apply forces to the thing in an analogous fashion, the thing might move around a bit, but it just won't be same!
Or, how about this: if you dissemble the car's motor, will it run? Will it amplify the motion of gas going into the engine, turning it into a far larger motion at the drive shaft? Same thing as dissembling a transistor -- the thing will longer turn a smaller current, flowing through the Base-Emitter junction, into a larger current from Collector to Emitter [or vice-versa].