Quisp brings up a great point:
If anything, the league ought to redesign the turnbuckle, so it isn't so hazardous.Damn right, Quisp. We've got legions of unemployed engineers in this country. There are legions more in the tundras of Canada. Why don't we try to create some solutions for this problem?
I've created a couple options for the NHL to consider below. All of these will require some measure of change for fans sitting along the glass. If the NHL is serious about reducing injuries, then they need to consider these options even at the expense of reduced high end seating.
1.) The ramp.
Instead of having a stanchion for the glass that begins immediately at the end of the bench, continue the glass down to the boards along the door to the bench. Pad the top of the ramp. Now, instead of slamming into a pole, the player will be hitting a ramp, slowing him down as his feet leave the ice and he drags along the upper padding. This method has the added advantage of reducing the number of pucks sent into play and thus reducing the total number of stoppages.
Negatives: Makes it harder for players to hop over the boards along the bench. Some reduction in glass protection for the spectators. My diagram shows one option, where a small triangle is removed from the glass. The other option is reducing glass height all around.
2.) The hinge.
Replace the bottom portion of glass with a hinged door. The glass is angled slightly (3-10 degrees) towards the bench. When a player collides with the door, he'll push it to an angle of up to 45 degrees, thus ramping himself back onto the ice. This is a cool design because you could, theoretically, automate a lot of its functionality. You could have it stay parallel to the boards until a proximity sensor detects an imminent collision, at which point it angles a few degrees so that the player does not contact the vertical portion of the door. Automation makes this method ideal, as being able to keep the glass parallel to the ice is important for gameplay. Automated or manual, though, this solution would work.
Negatives: You'd lose two, possibly three seats along the glass. The system is reliant on a shock absorber to cushion the player as he impacts the glass, and tuning this shock absorber to properly deflect with the disparate sizes of NHL players might prove difficult. The non-auto version will result in glass which is not parallel to the ice, inhibiting clears along the glass.
This is just a large sheet of gently curved glass which extends two or three feet behind the boards and continues out to meet the glass beyond the blue line. It's way better to run into curved glass than a vertical stanchion. Simple, effective, and thus unlikely to be considered.
Negatives: Lose some space on the bench, curved glass is more expensive/fragile than standard glass, not as effective as the two previous methods.
Alternatively, players could try and care about their colleagues and not intentionally ram them into the boards along the benches. But as we all know, that's as likely as NFL players agreeing to wear real helmets.