[entwife]

A brief review of a single Wikenigma article suggests that it doesn't cover known topics particularly well. The mechanism for color vision in humans is poorly understood. We do know that there are four ocular photoreceptors, and at least one extraocular photosensitive cell in humans. The Wikenigma article erroneously states that "human eyes have three different types of retinal cells (cone cells) which respond to light". Of the several human photoreceptors, it is accurate that cone cells respond to light intensities at which we perceive color, but these are not the only human photoreceptors.

[kvathupo]

It sucks to criticize, but I concur in the case of the site's article on neural networks [1]. They state that the successes of neural networks "defy mathematical understanding." Universal Approximation Theorem is sufficient; even if you give the benefit of the doubt that they refer to interpretability, there is active and fruitful research into NN saliency. (As an aside, I'm curious how such a quotation got published in PNAS?) It goes on to criticize NNs as not being biologically accurate, misunderstanding the purpose of NNs.

Perhaps this is due to the site relying heavily upon newspaper articles, as opposed to actual research.

[1] - https://wikenigma.org.uk/content/mathematics/neural_networks...

[mdp2021]

The stated aim being «to act as a Catalyst for Curiosity in a general sense – as well as trying to identify possible starting points for (re)cultivating interest in scientific, academic, and of course, armchair-based research [... also as] curiosity does tend to wane significantly as years go by», the ambition is not the most intent and practical.

https://wikenigma.org.uk/curators_rationale