Now, to be clear, I would be only too delighted if that was true. We need Labour to be defeated in this general election and we need to maintain a pro-independence majority among Scottish seats at Westminster. I have no time at all for the crazy notion that what pro-independence voters should be doing at the election is "getting rid of useless SNP MPs". However, delusionalism is not going to help SNP MPs retain their seats. We have to be clear-sighted about how serious this crisis is if we're to have any chance of resolving it.
The evidence that a significant minority of former SNP voters have moved to Labour is plain as day. Skier vaguely claims that SNP voters are going missing in the polls and that the weightings somehow show this, but in fact in the latest Survation poll, the unweighted number of respondents who recall voting SNP in 2019 is pretty close to being in line with the correct figure, and thus not much weighting has had to be applied. Once the turnout filter has been added, 69% of SNP voters from 2019 have kept faith with the party, while 21% have gone to Labour. By contrast, Labour have retained 90% of their 2019 voters and lost only 8% to the SNP (the latter group probably includes a fair few disillusioned Corbynistas).
Because far more people voted SNP in 2019 than voted Labour, the 21% of SNP voters who have since switched to Labour amounts to more than six times as many actual people as the 8% of Labour voters who have moved in the opposite direction. So it's not rocket science as to why the SNP lead has contracted in some polls, and disappeared altogether in others. Instead of pretending this hasn't happened (real votes in the Rutherglen by-election give the lie to that anyway), the SNP would be better advised to make the strategic changes required to win some of those voters back as a matter of urgency.