I'd be a bit cautious about using an online colour vision test. A lot depends on the type of monitor you have and its calibration. Paper Ishihara-type tests are very complex to print accurately and need about 7 or more different inks rather than the 4 used in ordinary CMYK printing.
The Ishihara plates most certainly do test colour vision, through testing the perception of patterns that you can only see if you have good colour vision (and a few where you can only see the pattern if you don't). It's useful 'cos it's cheap and if you pass it, you're very unlikely to have any deficit whatsoever with your colour vision. In the jargon, it has a very high negative predictive value. It doesn't tell you whether a colour vision deficit is serious enough to matter when flying, but it's a cheap way of telling the majority of people that they're in the clear. If we got rid of it and made everybody do the FM test (a very poor test, in my view) or Nagel anomaloscopy (takes a fair amount of skill to use accurately), it would probably add another £100 - £150 to the cost of a basic medical.
I'm quite sympathetic to the idea that people with defective colour vision could probably fly safely at night, but there are lots of points I could pick with the article that Peter posted.
For example:
There is no evidence that colour defective vision has ever been a factor in the cause of an accident in civil aviation, and as far as I can ascertain, in military aviation (the military are not as open to scrutiny as they ought to be).
Some of the issues that he mentions (e.g. cockpit light identification) could potentially be an issue during the day, but really what we're interested in here is night flying. If colour-anomalous people can't legally fly during the night in most places anyway, then we're unlikely to see many accidents at night that are due to defective colour vision.
People with good colour vision sometimes have midair collisions, but even if people with colour anomalous vision were much more likely to have midair collisions, in practice it would be impossible to find evidence for this. Simply put, it's unfair to ask people to prove the unprovable, and illogical to assume that because the unprovable hasn't been proved, it isn't true.
The author also seems to me to fail to properly address the issue of 'popout' in relation to collision avoidance. Popout is a phenomenon where your attention is drawn to something even if you weren't looking specifically for it. For example, if all the stars are white except for one red one, your eye will be drawn very strongly towards the red one - assuming you have normal colour vision.
As an example, imagine you're flying on a collision course towards a mast in a built-up environment full of other lights. Whilst it's quite correct that if you've identified the mast, parallax cues will tell you that you're going to hit it, if you haven't realised that you're flying towards a mast this cue isn't much good to you. And if it's lit by a fixed red light, it's likely to be much more apparent to a colour normal than a colour-anomalous pilott.
For people with severe protanopia, red lights can also appear much dimmer than for other people, which makes them less visible. I suspect this problem may worsen in the future if lighting is gradually changed from filtered incandescent to LED lighting, because red LEDs can produce a longer-wavelength red than incandescent bulbs.
I completely buy the argument that strobe lights probably pop out better than a red light, even for colour normal pilots. But the fact of the matter is that lots of obstructions are lit with fixed red lights, not strobes - perhaps because parallax is harder to estimate for strobes?
I've often wondered whether the solution could, in part, be to change aviation lighting slightly. I haven't thought about or looked into it in detail, but if we used a bluish-green light rather than a green, people with red-green deficits might be able to identify nav lights much more accurately (there could still be problems if looking directly at them). If we used a very deep orange rather than a long wavelength red, protanopes would find it considerably brighter. And if we put up at least one flashing beacon on radio masts, everybody would find them easier to detect.
Obviously this would take a while and cost a bit. In the meanwhile, why not simply accept that the collision risks of doing the night rating under radar, in good weather and along a known route with no obstructions should be miniscule. It would let lots of people get a CPL or restricted IR.
~~~~
I once went to a conference on colour vision in the USA. One of the exhibitors told me that in his state, colourblind people were not allowed to hold a commercial driving license 'which is part of the reason we have the safest roads in the world'.