Nullius and DannyHamilton are spot on.
It's sad that people are getting bamboozled by malicious disinformation on this subject.
The _exact_ same thing protects segwit outputs from being stolen by malicious miners as any other coin: Following the rules is part of what _defines_ mining. A miner that steals an output hasn't mined as far as nodes are concerned, their blocks will simply be ignored (and the peers relaying them eventually banned).
Segwit is no different from any other consensus rule in this respect-- other than some were introduced later than others, but many have been introduced over time.
We didn't see these same sorts of malicious FUD with P2SH though it was exactly the same-- I guess because back then felons hadn't figured out how to monetize that sort of confusion.
Indeed, the Bcash scammers finally found a way to get money out of it, BUT there are groups of people not selling a shitcoin, an ICO or any other monetizable product whatsoever, which also think segwit poses risks, for example I remember this log from MP that someone posted in the bitcoin classic thread:
thestringpuller: cause there is no way TRB will ever enforce segwit, so there is no way it can ever truly verify a segwit output was spent "legitimately"
mircea_popescu: thestringpuller the derp in question can just spend again normally and his coins will be visible.
mircea_popescu: trb has no notion of "coin history". nor should it. because taint is not a thing.
thestringpuller: gotcha. just never have anyone spend to your from a fake address.
mircea_popescu: moreover, there ISNT, in general, and for very good reasons, a way to verify segwit crapolade.
mircea_popescu: which is what it aims to be, a sort of "let's dao bitcoin"
thestringpuller: TMSR rule of thumb: "Never accept non standard transactions" ?
mircea_popescu: pretty much.
thestringpuller: gotcha. thanks for clarity. mod6 ^^^ nvm question has been answered.
mircea_popescu: but this said, yes it is deeply irresponsible for anyone to use prb clients. this doesn't just mean 13, or 12, or 10. ANY of them.
mircea_popescu: they keep adding shit, but it's been shitsoup for years now.
mod6: thestringpuller: np.
mod6: if we needed to add code everytime these gnomes comeup with a new crapolade, that's all we'd ever be doing.
thestringpuller: i just don't want them to add something in a "fork" that allows TMSR to get scammed.
thestringpuller: but I think at that point BTC is dead
mircea_popescu: and in other lulz, ro chicks trying hard :http://fabulousmuses.net/2016/06/marina-yachting-summer-trends.html
mircea_popescu: mod6 the main concern is that their bullshit will "accidentally" start fabricating coins.
mircea_popescu: nevertheless, we're not making the mistake of introducing coin taint, under any name.
BingoBoingo: thestringpuller: From what I understand Segwit to a normal 1xxx adress requires a signature in the blockchain so when segwit stops being miner enforced the recieved transaction would still be, even though it came from freemoneyshitsoup.
thestringpuller: BingoBoingo: AHA!
mircea_popescu: yeah, except the next stop is for shitsoup to put out more coins than it got in.
thestringpuller: elaborate?
mircea_popescu: you been watching the dao thing ?
mod6: <+mircea_popescu> nevertheless, we're not making the mistake of introducing coin taint, under any name. << totally agree.
mircea_popescu: segwit is EXACTLY "dao for bitcoin".
Im not too familiar with this but apparently MP got a lot of bitcoins, and these guys are not trying to scam anyone with shitcoins (forks included) and as far as I understand they are trying to do what's best for bitcoin, so I value their opinion on the matter. I would like to know what you think and why there are big discrepancies with Core, because these must be real technical reasons, since again, they aren't selling their own scamtoken, as Roger and co do.