This is the Neanderthal method of gauge calibration, but it is worth a try:
Just make a mental note of where the needle sits when at rest. Then take your finger tip and force it past it's zero point, in the opposite to the direction of the error - in your case that would be CCW. Try forcing it maybe an inch below zero (told you it was crude) and let it go.
Take a test drive and see if it tracks more accurately. You won't notice the difference at low speeds - take it up to 65 -70 MPH and compare the reading with your GPS. If it helped, try pushing it another inch counter clockwise until you get a fairly close reading as compared to the GPS.
Be aware, the dampening force of the spring is not linear. It offers very little resistance to the rotation of the output shaft at lower speeds. It should come to rest a little bit lower than before you twisted it backwards but it will not exhibit that degree of "depression" across the entire scale.
So, you should not see a big change at low speeds. But then, as the shaft/needle begins to move up the scale, the spring begins to offer greater opposing force. The higher the needle climbs, the more the spring tends to limit the motion.
With any luck you will have reset the spring to a tension that will hold the needle around the 65 MPH mark when you are actually going that speed.
When all else fails, just pull the needle off and set it back on the shaft the amount CCW that corresponds to your 10 MPH error. You won't show any speed until you hit 10 MPH but you will be accurate at highway speeds.