The scenes shown in the example link you provided are perfect examples of what Paul suggests, the scenes are dark because they properly exposed and the sample of the one appearing brighter might be over exposing based on the meter's goal. By learning what the meter is telling you, you have all the information needed to get the image display your wish. The purpose of the meter is misunderstood by most photographers of modern cameras because when the range of light is wide, the auto exposure modes and our expectations are very close. When there is a smaller range between darkest and lightest, the metering remains consistent, but our expectations are off. The scene IS 18-20% middle grey so it is showing what it is designed to show you.
The camera has a lot of built in biases that are selected based on metering mode and color components based on analysis of real world images in Matrix mode, so its guesses are pretty close to how people expect the scene to look like to them that works well in wide DR scenes or with wide chroma range. The success in guessing your expectations in Matrix mode is pretty good but the middle grey definition of mid point of scene brightness still rules.
The brightest possible scene element that can be captured and the darkest element that can be captured depends on ISO, the widest range is at base ISO and the very smallest range that does not exceed the DR of the camera is at highest ISO so wide DR scene capture is crippled by the current fad to expect very high ISO performance.
Does the camera generate images that look more as expected when there are elements both bright and dark where the middle point between those extremes is brighter? I suspect in scenes with a brighter mid DR point look as you expect.
When a scene is dominated by a level of light, it is going to be captured as middle grey, whatever that dominate level is. Take a photo at the beach on a bright sunny day and you would expect to see it as bright in a photo because of our experience of sunny beaches but the camera has no idea what beach brightness is and only seeks the mid point between darkest and lightest. If EVERYTHING is very bright, the mid point is bright to you, knowing what you are looking at but to the camera that bright mid point is the mid point between deepest black and brightest full well illumination so decreases sensitivity so the mid point is captured as 18% grey. The example to show what middle grey is can be seen with mixing paint. The blackest paint you can find and the brightest white you can find when mixed together in some ratio so that it appears to be a mid point in brightness between that two extremes turns out to be adding 18% of the volume of black to the white paint produces a mid level of illumination. Or adding 82% volume of white paint to the jet black paint produces that same mid point between the extremes of brightest to darkest.
In your example photos the meter says that the brightest part of that scene and the darkest part of that scene are close together in absolute illumination so a dull grey is that that mid point it wants to capture. If there were very bright elements and very dark elements introduced to that scene it would still see middle grey as dull grey but your eye would see the wide DR of the scene and think it was normal.
Shoot a scene that is darker, where the deepest black and the lightest parts of the scene are still dark, the meter thinks the mid point between those two extremes is the mid point to capture as the mid point of the camera's brightest and darkest recording capability, which is a lot brighter than your eye sees it. So it appears over exposed in the capture.. Take a photo of a groom in a wedding where his Tux is the dominant element under the focus point. You look at the monitor and are disappointed, his tux looks dull grey. Next you take a photo of the bride where her very bright white dress is the dominate level under the focus point and the capture is a dull grey, under exposed to your eye. In both of these cases, a wedding photographer would know to get black elements black looking, he has to dial in negative exposure compensation,1.5-2 stops as a starting point. With the bride, be dials in positive exposure compensation. But if he shot them both together in Matrix mode, the mid point between brightest and darkest would be pretty much how you would see it in real life and the dress would look a great deal brighter than the dark tux.
When shooting snow scenes you know from experience, the photos look grey, not bright white, for the very same reason. Dialing in 2 stops of positive exposure compensation results in white snow unless there is a large dark object in the scene, in which case the mid point would be captured more realistically to your expectations. You know it is snow and expect it to look white but the camera has no idea what it is supposed to look like to you. If bright white is the only element in the scene, it is going to be exposed as 18% grey.
Knowing the foregoing puts you into full control of what the scene is captured like as perceived by humans. When you understand what the camera sees and what you see being different, it allows you to predictably compensate to get what you expect. That is when you start becoming a photographer instead of a picture taker.