what do you think of his theory?
His account of the sequence of events / activities is intermittently sound. However, the specifics of his theory are based on references to geographical / map features (especially 'hills') which don't exist and call into question his ability to understand a topographic map.
The naming of 'hills' is based on taking the elevation of a height's peak as the designator for the overall hill / mountain. Neither his 'Hill 805' nor his 'HIll 663' exist as peaks.
'Hill 805' may be a typo intended to be 'Peak 835' - the intermediate rise or hump between Kholat-Syakhylor and the flanking peak to the east. Alternatively and more probably it's an allusion to 'Peak 905' - the outlying peak to the east. Wilkins' murky explanation of a navigational error seems to suggest the 'Peak 805' Dyatlov erred about was in fact 'Peak 905'.
'Peak 835' is hardly 20 - 30 meters higher than the saddle (top elevation) of the pass itself, but it would have been a geodetic reference point on the topographic map available in 1959. It's certainly marked on the later topo maps.
There is no 'Hill 663'. There is a reference landmark on the slope leading up the south side of the ridge the party climbed marked at elevation 663 (m). This is not a peak at all - it's an arbitrarily designated spot on the side of a slope. I'm not sure about the map symbology for this spot, but I suspect it's a ruin or other landmark. I'm also uncertain whether this spot on the slope was marked on the map(s) available in 1959.
There is no 'Hill 611', either ... The '611' designation is a reference point for the *lowest* elevation in a vale / valley on the north side of the pass - the very vale into which the doomed party fled and died. In other words, it's the exact opposite of a 'hill'. According to diary entries the original plan was to cross the pass, set up camp and possibly their cache in the vale beyond (i.e., the '611' area), and approach Ortoten using the 'low road' strategy (i.e., staying in the watershed's forested valleys until a final climb up Ortoten).
Notice that Wilkins' theory is partially predicated on T-B and Zolo being prepared to get oriented to the location of Hill 611 if visibility ever improved overnight. There was no Hill 611 toward which they would seek to obtain such a bearing. Moreover, no such bearing was necessary in the first place. Map point '611' was the valley into which they'd flee and die, and all they had to do to find it was to follow the slope downward from their tent site - precisely what they did (presumably during the night, by standard interpretations). Dyatlov was carrying topo maps of the area, and he was experienced enough to read them.
Pitching camp high up on the flank of Kholat-Syakhylor suggests either (a) they had to abandon their original plan to continue downhill into the valley on the north side or (b) they'd decided to stay at the higher elevation and approach Ortoten using the 'high road' strategy typically employed in the summer months. Option (a) had still been in effect the preceding day, because they attempted to scale the pass before establishing their cache. The conditions and approaching sunset forced them to retreat back into the river valley from which they'd come.
Option (b) must have been decided that night or early the following morning, because they set up their cache before making their second attempt to climb the pass.
To the extent the high road strategy would minimize subsequent climbing I agree with Wilkins that economizing time and effort was or could well have been a factor in whatever their decision making process was.
Except for the geographic reference bungling and blaming everything on compass-based navigational issues most all the rest of Wilkins' exposition repeats stuff already known and mentioned here.
There are other errors in his exposition - most particularly mis-attributions of what some of the images represent and misstatements of facts in his dramatization of the search party's experiences.
Even ignoring these other miscellaneous errors I have to give Wilkins' analysis low marks for basing his theory on a clearly mistaken interpretation of the basic landscape.