He does, but the writeoff can't be bigger than what he invested.
Except when it can. It's often possible to do several investments at roughly the same time (within a few years of each other), lose massively on one and write off more than you invested on that one as long as you make a profit on the others and the profit is larger than what you want to write off.
I find it fascinating that we have no way of telling if a user is human just by looking at their signal.
All the meaning we attribute to a signal depends entirely on something that is not detectable from the signal itself itself: wether an actual human is behind it.
This phenomenon intersects with philosophy, cognitive science, and information theory, and the best solution corporate automata come up with is "just let them pay for the usage".
At some point we’ll realize it doesn’t matter if there’s a human behind. Say if there is a human behind claiming to be 20 year old girl from Russia mad in love with you, but it’s a Nigerian scam run by a 50 year old man, does the fact he’s a human help you? It matters in terms of voting and forming overall opinion. But I’d say it doesn’t matter for individual interactions. It can always be a scam.
It's not too hard to build a decent detector. At one of my previous gigs (a niche social network) we trained a fairly OK NN to detect all sorts of undesired behaviours to flag content/account for moderation. We didn't even need that much data and training time for it to get OK.
In case of X its scale is the issue. Running the detector for every message, or even for every posting account once a month might be very expensive. This might be the primary reasoning behind the deliberation: make bots a little bit more expensive and finance the detector operation.
Likely not. Here's a paper that provides a decent argument that you just can't be sure that two Twitter accounts represent two different real people.
«We posit the Ghost Trilemma, that there are three key properties of identity -- sentience, location, and uniqueness -- that cannot be simultaneously verified in a fully-decentralized setting. […] We sketch a proof of this trilemma and outline options […]» https://cs.paperswithcode.com/paper/sok-the-ghost-trilemma
I'm not even sure this will help much. Yes, in addition to AWS account bot farms now would have to pay X per-bot but this is hardly an issue for state-sponsored bot-farms or bit dark net operators.
With that it's going to push out lots of human users who either can't afford to pay or don't see enough value to justify the price (e.g. casual/occasional users). And with outflow of audience value diminishes for users who might be willing to pay, pushing some of them over the edge.
I wonder whether this would actually increase bot-to-human ratio.
My prediction if would get implemented it would last a week. Like with login-wall last time numbers would plummet too much.
I think it very much could help. As it stands there is a minimal cost per bot account, if it gets banned you try again.
With a monthly subscription you'd be out of pocket for a month, even if your bot is banned the next day. That's terrible ROI for a spammer.
Furthermore getting a supply of "plausible" cards (corresponding to the account's/IPs location) that would pass fraud checks is not as cheap as phone numbers either.
It will be effective. Large state-sponsored actors will no doubt get through it, but it still shrinks the amount of spam significantly.
I feel like there are a few assumptions in there that might not be true.
First, I don't think there ever will be a limitation of 1 (or even small n) account per card. This excludes big corporate clients, media networks, etc. who have legitimate reason to have more than one account but would still pay from the same account. I'd bet a buck these are the accounts that most likely to actually pay for X. It would be unwise to gate them out.
Second, if detecting bots would've been as easy as taking only a day (even a few days) worth of posting then there would never be a need to paywall the whole of X. I believe it's because it's hard (either objectively hard, or just expensive for X) to detect bots Musk considers such a disruptive measure.
Third, cards are probably not a good verification mechanism. More so if there's not card-to-account limitation. Stolen cards don't have to be used directly. Money can be laundered elsewhere and then used with a good card. It is an obstacle but probably only for small/ad-hoc operations. Big bot nets will hardly be impacted by this.
So many companies are so nervous about having any user-generated liability that KYC is on steroids. Lots of services want to tie online accounts to physical identities (and, hopefully, those identities' bank accounts). Meta already pushed for drivers' licenses. Quite a few sites use Veriff or other services that match the biometrics of IDs with selfies/videos. Passport might be a bit too stringent but I can definitely see Musk wanting a document that conveys proof of citizenship which is a whole other mess, I hope all the non-US countries on the planet enjoy him having a cobbled database full of their citizens' identifying information.
As I don't believe there is too many users who will pay for it and this will destroy relevance of it for everyone who is paying...