Although - unlikely as it seems, you may be onto something here.
There is a catch, of course.
The closest thing I can find to your proposal is a reference in which benzyl alcohol, acetaldehyde (ethanal) and methanol are reacted together, and undergo the following reaction. The product is 1-methoxy-1-phenyl-2-propanol:
Molecule: (https://www.the-hive.ws/forum/faq.pl?Cat=#applet)
("c1ccccc1CO.CC(=O).CO>>c1ccccc1C(OC)C(O)C")
Epoxides, diols, O-acetylated diols, and probably O-alkylated diols can be isomerised to P2Ps with sulfuric acid. However the abstract of the article for this transformation makes reference to chromium tricarbonyl, which makes an otherwise very interesting idea much less so.
Still, the reference is:
Elaboration of alpha-substituted benzyl ethers and sulphides by suppression of the Wittig and related rearrangements
Stephen G. Davies, Nicholas J. Holman, Charles A. Laughton, Bryan E. Mobbs
J. Chem. Soc., Chem. Commun., 1983
[No DOI found]
Abstract
Co-ordination of benzyl alkyl ethers and sulphides to chromium tricarbonyl allows a-substitution via the corresponding a-carbanions to be achieved by suppression of the Wittig and related rearrangements.
If it did turn out to be a viable route, it would be quite an elegant pathway to meth, via oxidation of the alcohol (the other one is protected as an ether; Jones reagent seems selective for this) to the O-methyl phenylacetylcarbinol (O-methyl d,l-PAC), followed by standard reductive amination. An HI reduction would remove the -OMe, instead of the usual -OH, giving methamphetamine. See Post 344446 (missing)
(Rhodium: "Hypophosphorous acid¯iodine: a novel reducing ...", Stimulants) for an example in which -OMe is reduced by an HI-based system.