Starting with a higher-res scan and then resizing to a smaller resolution might still result in better quality.
I'm not sure if this is a fair comparison, but years ago a DVD review site took the Blu-Ray release of Lord of the Rings, ripped and processed it to DVD size, and compared the output to the original DVDs (which used to be considered one of the best quality DVD releases when they came out). The quality difference was huge.
Starting larger than the target delivery has always been a known thing. Hell, they used to shoot 35mm for SD. Down scaling in and of itself is a way of reducing noise, as the noise averages out in the scaling process. Of course an HD image scaled down to an SD will look better than a DVD compressed from an SD source. Even if the SD was from a DigiBeta master, an HD captured from a HDCamSR or straight from a digital file format will just have so much more detail and information. The bit depths alone in HD sources can make a difference. The HD format allows for so much more information. Why this is surprising is the real surprise to me. Not from you as much as the people attempting the actual test. Sounds like something the readers of doom9 might have done for the clout.
We don't necessarily know what the original digital transfer method + processing pipeline was for the DVDs. Comparing them to the Blu-Ray releases lets us make better informed guesses.
I'm not sure if this is a fair comparison, but years ago a DVD review site took the Blu-Ray release of Lord of the Rings, ripped and processed it to DVD size, and compared the output to the original DVDs (which used to be considered one of the best quality DVD releases when they came out). The quality difference was huge.