gattsuru
2hr ago
This seems reasonable. My threat model is a little broader than yours -- the question of 'what happens if something I've already done and isn't controversial today becomes a felony', possibly without me knowing or having prior notice, is a little more prominent in my mind, as a not-straight furry who follows a lot of CTRLPew stuff -- but for a lot of people (and, honestly, even my own use case) there's reasonable questions about where this falls into paranoia.
gattsuru
2hr ago
There's been more than one cloud provider data failure, but yes, it's absolutely true that the average cloud provider (even including the bad ones!) on average are better than the average end user (even excluding the very bad ones).
My argument is more against the framework where the Cloud solves the problem, either entirely or to such a degree that end users don't really need to consider it. That might seem like a strawman -- everybody talks versioning and backup! -- but Actual Professional Standards often difference between centralized and decentralized backup approaches, people colloquially treat the 3-2-1 rule as solved by using different regions or services on a single provider (or worse a locally synced version) for 'media', and I've seen no small number of mid-sized deployments that have bought into it hook, line, and sinker. AWS in particular has a whole spectrum of (weirdly price and provisioned) services Just For being your all-in-one backup solution. There have been IT people responding to this incident as though Unisuper recovered its data thanks to help from Google, rather than a secondary provider.
That's not a Cloud-specific problem. People could -- and people did and do! -- leave all your backup drives plugged in and spinning at the same computer with your live copy, running in your personal server closet. But you could (and would) get slapped in any credible audit.
I think Cloud makes this particularly dangerous because the actual processes and ownership are often obscured or multilayered, at the same time that an increasingly few baskets have increasingly large portions of the eggs. As a business, if you don't investigate close enough, you can find out that your 'multiregion' is really just two sides of the same data center. As an enterprise user, if you don't watch your e-mails, you can find that your data is gone even if the actual hardware owner knows what they were doing. ((And along those lines, the "accidental" in "accidental loss" is some work, here: Google either lost 0% or 100% of Album Archive, depending on how you look at it.)) As a personal user, you can find out that wildly-different subscriptions all use the same (sometimes intermediate) provider.
I guess my point isn't to say that Cloud is bad, but that the Cloud isn't magic, even if it looks close enough for most people that they've stopped thinking about the matter. If you're going to use a cloud-style approach, a single account at a single provider can't be the end-all be-all for any seriously critical data. It's better than just storing your one live copy -- but better than nothing is not a complete and balanced solution. And this incident should put anyone in that sphere on notice.
((And for the broader theme for this roundup, sleep-walking into disaster because someone else will solve your problems for you runs into problems, even at a level where the 'someone else' is either Google or the Aussie regulatory system.))
gattsuru
4hr ago
How is it difficult? I regularly takeout photos at original resolution and then compress the cloud copies.
It's not unusually bad, either by Google standards or by the Cloud in general*, but Takeout has almost no tech support, can take arbitrary amounts of time to create the archive files, throttle(d) after daily download bandwidth limits that are often far less than the typical Takeout, can miss data, and is weirdly inconsistent between Takeouts. It's very far from the typical Photos experience, both in terms of user experience and in terms of literally not being part of the Photos UI.
- compare Ring, where bulk downloads of more than fifteen videos at a time requires either a warrant or the use of a half-broken python script.
gattsuru
21hr ago
Even contemporaneously with the release of the story, Yudkowsky was complaining at length that people read both Harry and Quirrel as both far more correct and far more competent than they actually were, whether not noticing their failures or overstating their accomplishments. Reality ended up pushing that even further, for Harry -- the first twenty chapters are filled with a lot of pop social science that was iffy to start with and didn't really survive the replication crisis -- but there are other errors that I think were intentional, even fairly early on.
gattsuru
22hr ago
One argument is dimensions. The difference between Falcon9 and a Starship to launch a cubesat is just economics, but the difference between the two to launch a fuckoff big mirror, not having to fold it up into a million pieces pretty much allows an entirely different and better design philosophy.
There's also some reliability arguments in favor of Starship's liquid methane fueling approach over the Falcon9's kerosene-fueled approach, especially given recent instability in fuel markets.
Presidential Ballot Access: Ohio Edition
As of today, a state is currently set to only have one of the two major parties candidates for President on the ballot, but it's not the one you might expect.
Ohio law requires that presidential candidates be certified – that is, the state must be notified that presidential candidates have been officially nominated – 90 days before the general election in order to get on the ballot. That is the earliest deadline of any state.
But the Democratic National Convention that will formally nominate Biden won’t open until nearly two weeks after Ohio’s Aug. 7 deadline. The Republican National Convention will wrap up nearly three weeks before the deadline, so Donald Trump won’t have a problem getting on the ballot...
Ohio laws generally take effect 90 days after passage. So a change to the deadline had to pass by May 9, but the Legislature wound up doing nothing.
To be fair, this law has existed since 2010, albeit with a couple past temporary exceptions (probably a compromise number downstream of a 2006 court case over a third-party candidate 120-day deadline; see references to Blackwell and progeny here). and the Ohio legislature (majority Republican) has done nothing less out of explicit desire to screw over democracy and more because the Ohio GOP's House and Senate are fighting each other, and a GOP interest in getting some sort of Red Tribe value out of it (the closest bill, HB114, also bans some foreign contributions to ballot initiatives, for about the reasons you'd expect). DeWine, the (Republican) governor, is pressing pretty hard to find some sort of solution, whether that ends up an emergency legislative fix, hoping the courts can and will step in, or a more dubious executive branch intervention. And it's not like there's any plausible situation where Ohio would be the turning point for the 2024 election.
((There's some theories that Ohio Dems are trying to bolster Sherrod Brown's chances, though I don't think that's very likely or even particularly coherent.))
To be less charitable, nice motive, still excluding a major political party from the ballot. It's not going to be doing any wonders for society, and as we get closer to the election, the available options, whether taken or merely proposed, will only make the mess clearer. The current planned resolution looks to be a 'virtual Democratic National Convention call', officially nominating Biden before the actual Chicago disaster convention. Hopefully, that's enough of an excuse for the Ohio legislative special session to also clean things up, but more likely the virtual call gets counted as close enough for Ohio law (less optimistically, it leaves no one able to challenge it). But the whole thing has just been a parade of one group after another absolutely certain that someone else will fix the problem that they're rolling directly into, and 'it wasn't my fault' is an awful epitaph.
Fauci et All Foiling FOIA
One email refers to a “private gmail” supposedly used by Fauci. Morens also referred others to a “secret back channel” for communicating about certain issues. He also frequently directed others to message him on his personal Gmail account to avoid FOIA requests. Morens also noted he had “learned from our foia lady here how to make emails disappear after I am foia’d.” That individual, who he identified as Marg Moore, “also hates FOIAs.”
The emails also reveal Morens made a series of crude comments about women, female coworkers and his drinking habits which indicate he “is not qualified to hold a position of public trust,” the committee wrote. Morens testified to the subcommittee Wednesday that he didn't remember if he used a personal email to conduct government business, but conceded it was "wrong" if he did so.
None of this is particularly surprising, from a cynic's perspective -- government employees fucking with FOIA requirements is a day ending in Y. There's a steelman where certain scientists involved in climate change research were getting spammed with so many duplicative FOIA requests that it edged on harassment, though given neither Morens nor Fauci every worked in environmental stuff that that's leafspring-grade steel.
There's no serious chance of serious punishment, here: Morens was already investigated and found not guilty of anything that the NIH cared about, and if anyone has problems with him lying to Congress, well, there's some fun legal realism questions about whether the law is the statute or the enforcement, but the enforcement still comes from one place.
On the other hand, it does seem enough to have pissed off no small number of partisans aligned to that One Place, if, cynically, more in the sense that Morens et all got caught. HHS is at least moving against EcoHealth Alliance. If you were to ask what one would consider a good sign, well, there's certainly end results that could point to people taking this seriously.
Title VII Religious Freedom in California
Another day, another VanDyke dissent:
In its stubborn insistence on ruling against Chief Hittle, the panel has twisted the record into knots and badly misstated Title VII law. Its decision (1) abdicates its responsibility to read the record in the light most favorable to Hittle at the summary judgment stage; (2) allows employers to escape liability for repeating discriminatory remarks simply by hiding behind those who say them first; and (3) mangles Title VII’s “motivating factor” analysis.
Perhaps most glaringly, its original opinion also incorrectly heightened the showing a plaintiff is required to make to demonstrate disparate treatment. In the panel’s view, Hittle bore the burden of showing that the City’s discriminatory conduct was “motivated by religious hostility,” notwithstanding the Supreme Court’s admonition that such a plaintiff need only show he was “intentionally treat[ed] … worse because of” a protected characteristic, Bostock v. Clayton County.
Recognizing at least this last mistake, the panel’s amended opinion retires its former use of the word “hostility,” replacing it with the more accurate (but less specific) “discriminatory animus.” Not only do those changes not fully fix the original opinion’s legal errors, but they also put the panel, which apparently remains as determined as ever to rule against Hittle, in a pickle.
Notwithstanding its many other errors, the original opinion correctly acknowledged that the “gravamen” of the “notice terminating Hittle was the religious nature of the leadership event.” But if attendance at a religious event was the “gravamen” of the firing and Hittle need only show that he was “intentionally treat[ed] … worse because of” religion, Bostock, 590 U.S. at 658, it would seem the panel would have no choice but to reverse its previous decision in favor of the City.
But it won’t.
We've had discussions here about a more expansive framework for discrimination, but this seems the punchline. Title VII has prohibited religious discrimination in hiring and firing of employees for sixty years, and while the exact borders of the doctrine have ebbed and flowed from one jurisdiction to the next, or as courts have pushed at the very edges, trying to bring them to these style of cases has been an expensive shitshow, where even the lucky winners spend decades for minimal defenses. Even defining this class of cases is a mess: I want to say discrimination against the 'majority' (but anti-woman discrimination is clearly covered!), or 'non-subaltern' (but trite agnosticism is protected, he says as a trite agnostic), rather than just Groups Progressives Want To Discriminate Against, and that's kinda the problem.
However, that punchline's also a bit of a repeat: not just that VanDyke is writing his dissent at an en banc appeal that had no chance of victory, or that the original opinion had to be edited to not be hilariously wrong, but that Kennedy went to and was decided by SCOTUS almost a year before the original appeal decision came from the 9th Circuit. The facts in Hittle are different, sure -- Stockton alleges, not very credibly, that they 'really' fired Hittle for endorsing a consulting business and for not disclosing closeness to a union president -- but the courts were (supposedly) not yet deciding facts, but merely the motion for summary judgement.
Instead of motions for summary judgement focusing on questions of law, various balancing tests and excuses can fall into play where judges don't like the plaintiff's perspective. Instead of protecting Hittle against employers that were outraged by his place in a complaint-named 'Christian coalition', the law in California now holds that there is a "legitimate concern that the City could violate constitutional prohibitions and face liability if it is seen to engage in favoritism with certain employees because they happen to be members of a particular religion." Sure, that anti-endorsement test had been explicitly rejected contemporaneously to and previously by SCOTUS, but SCOTUS "can't catch 'em all", and increasingly doesn't seem interested in trying.
That failure mode isn't and wasn't inevitable: despite my expectations, Fulton hasn't come back to the courts (yet). But it's a problem that haunts any attempt at legislative or executive branch 'fixes'.
((At a more concrete level, Hittle was fired in October 2011, at a time where he was nearing age 50. The unusual length of the court case here reflects Stockton's bankruptcy rather than overt malfeasance specific to him, but it still means he's in his mid-60s today. Even should, SCOTUS hear this case, overturn it, remand with direct instructions, no further interlocutory appeals or weirdness occur, and the trial occur speedily, he might see a court room on the facts before he self-moots by old age, but probably not before he sees his 70th birthday.))
An Appeal to Heaven
Compare February and to May. Diff, context. Also see here, and here.
The Cloud is Someone Else's (Broken) Computer
Unisuper is an Australian superannuation fund, which is close enough to a psuedo-mandatory version of American retirement funds. AshLael might know the more specific differences. It has 600k members, about one in fifty Australians, with over 125 billion AUD (~88 billion USD) funds under management. It also fell off the internet on May 5th, only restoring full functionality May 20th, allegedly as a result of a 'one-in-a-million' bug in Google Cloud services dropping both the main Unisuper database and all Google Cloud backups.
UniSuper had duplication in two geographies as a protection against outages and loss. However, when the deletion of UniSuper’s Private Cloud subscription occurred, it caused deletion across both of these geographies.
Restoring UniSuper’s Private Cloud instance has called for an incredible amount of focus, effort, and partnership between our teams to enable an extensive recovery of all the core systems. The dedication and collaboration between UniSuper and Google Cloud has led to an extensive recovery of our Private Cloud which includes hundreds of virtual machines, databases and applications.
UniSuper had backups in place with an additional service provider. These backups have minimised data loss, and significantly improved the ability of UniSuper and Google Cloud to complete the restoration.
This is a little weird, and not just for having an actual benefit from multicloud. Google Cloud Platform doesn't have the best reputation, but 'keeping multiple copies of long-standing data' is one of those things cloud providers are supposed to excel at, and having first disclosure come through the client rather than the cloud provider is a decision that Google Cloud didn't have to make. There are even arguments, a la Patio11-style, that part of what a client Unisuper's size is buying from a cloud provider is to have a name to fall on a sword. This has lead to no small number of people reading tea leaves to conclude that the fault 'really' reflected an error by Unisuper (or a separate smaller contractor) making a configuration mistake. Unisuper was migrating from VMWare, which has its own mess, and is exactly the sort of situation you would see greater vulnerability to client developer error. That still wouldn't be great for Google, since most cloud providers at that scale claim a lot of safety checks and emergency backups, but I could understand if they just failed to idiot-proof every service.
Nope :
During the initial deployment of a Google Cloud VMware Engine (GCVE) Private Cloud for the customer using an internal tool, there was an inadvertent misconfiguration of the GCVE service by Google operators due to leaving a parameter blank. This had the unintended and then unknown consequence of defaulting the customer’s GCVE Private Cloud to a fixed term, with automatic deletion at the end of that period. The incident trigger and the downstream system behavior have both been corrected to ensure that this cannot happen again.
We probably won't get a full breakdown until the Aussie regulatory agency finishes an investigation (if then), so there may always be more to the story, and a lot of fun questions about what, if any, data was out-of-date or lost from the backup. But this is pretty damning for Google, as things stand.
On one hand, this probably is a one-in-a-million bug, and readily closed. On the other hand, as anyone with network engineering or statistics or X-Com background can tell you, one-in-a-million means a lot less than the naive expectation, and Google Cloud Platform has an estimated just under a half-million business customers, and this is an embarrassing bug.
Does that mean that they're got it out of their system for another million customers? Or that this is just the first time it happened and was big enough a deal to make the news?
((Okay, the real answer is I'm being pedantic and reading too much out of a turn of phrase.))
The bigger problem is that Cloud has long been sold as The Professionals Doing It Right. That was always a little more true at the margins than the center. As bad as rando small businesses trying to maintain a Gitlab instance can get (and it can get bad: even by web software standards, it's a masochist's tool), no small number of seemingly-legit smaller cloud providers have gone belly up, wildly revised their offerings, or just plain disappeared. It's quite possible that Unisuper moved to Google Cloud in part because the Broadcom buyout of VMWare (only finalized in December 2023, but after 18 months of regulatory review) raised concerns that they'd start lopping down product offerings.
(But to move to Google?)
It's increasingly tenuous, here, though. Unisuper are not fresh college grads confused by the difference between a RAID and a backup, but a massive company that maintains many sites as a matter of course. Google is one of the Big Three when it comes to cloud provisioning. One can imagine counterfactuals where a self-run or classically-hosted Unisuper herped the derp, but the factual here makes them look like the competent ones. And that's not alone.
Which would be one thing if Unisuper were the only people pressed into cloud services.
How's everyone feel about OneDrive integration in Windows, or Google and Apple cloud in their phones?
Eh, I'm familiar with it, but I dunno if it's as good as it sounds.
Texas and Missouri have been poking at Media Matters over the Nov 2023 report claiming Twitter was running a bunch of ads alongside actual-nazi content. Twitter claims that, to do so, Media Matters had to go through some impressive hoops -- using older accounts following only the actualfa and major brands, generating massive numbers of ad views to cherry pick the bad combinations -- none of which were presented in the initial report.
Texas started their investigation just days after, but Media Matters was able to file in DC and get a preliminary injunction. Missouri started about a month later, currently trying to use jurisdictional stuff to avoid that same DC court from slapping them with a TRO.
I'm... not really happy with pretty much any part of this.
Both state investigations are premised, to the extent that they're premised at all, under the theory that MMFA defrauded its donors with this report. I can certainly believe that MMFA's writers were cackling with glee as they completely manufactured ad combinations -- this is the same corp that promoted a holocaust denier's false claims that a Texas State Senator had given him an interview until they got called out -- and maybe there's some smoking gun of them doing so for the donor revenue rather than hating Elon Musk, but I don't think anyone donating to MMFA would have held back their quarters because they were too mean or manipulative. At best, the state AGs are just using whatever law comes to hand, and while it's funny to see turnabout as fair play after Remington v Soto, it's not great for society. I don't even think it'll be effective (cfe Paxton injunction): these laws aren't swords that cut both directions, and neither is process-as-punishment.
On the other side, it's hard to see this as even-handed application of the law against Texas or Missouri, either. We have a major case before SCOTUS right now in NRA v Vullo about whether actual enforcement of a law in a way clearly meant to retaliate against and quiet political speech 'counts', and it's not some clear and obvious matter how SCOTUS will roll! Here, instead, simple investigation of violations of a facially neutral law are clearly wrong. The DC Court even has to go out of its way to separate this case from other cases involving Twitter and Paxton to explain why suddenly the normally-difficult preliminary injunction is both necessary and obvious.
There's ways to distinguish that previous case, and it's not even a Blue-Tribe friendly decision despite the 9th Circuit being involved, but hard to come up with a distinction broader than "2021 Twitter should have found someone willing to make a more specific declaration". It's very hard to see a distinction from the more common politically-driven investigations.
More immediately, I'm also skeptical that this lawsuit is the sole thing driving the firings at MMFA. Any legal case is expensive, and juggling multiple cases is even harder, but the Texas and Missouri lawsuits are early enough that they're not driving anywhere near the billable hours of Musk's civil suit. Compared to the normal ebb and flow of financial spending (MMFA gets almost all its contributions through Bonner, but the last leak points to a small number of major donors, which tends to be mercurial), there are a lot of other explanations.
gattsuru
5d ago
Even decent doctors doing 'normal' stuff can have surprising gaps, if they're struggling with combinatorics or recall -- outside of test reqs, it's one of the big arguments against the residency-as-hazing, because being that tired makes you stupid. This seems like a joke, but it's also a joke that I know three people in meatspace with similar stories specific to Crohn's. I've personally been given a combination of prescriptions that, about six months earlier, had received a black-box label about risk of horrible wasting cancers in my demographics.
((That said, I'm skeptical on both the naive HBD take, and also on the data here being completely causative, though I expect the base problem of 'prioritizing everything over ability' is bad enough even in a perfectly blank slate world.))
gattsuru
6d ago
Yeah, that's a limiting factor. I'm still really proud of this, and people did say variants of 'this is great', but the only real possible responses were to either get on a soap box or prove me wrong, and that limits engagement. This worked better in the sense that giving several different perspectives on a social problem means that there's something for everybody to disagree with, but it did so at the cost of wordswordswordsing.
I don't think engagement is the end-all be-all, but it does put a thumb of any measurement of the system here.
A division between Stand Your Ground and Duty to Retreat philosophies exists, but in addition to the lines being a lot blurrier than this summary (eg, even a lot of SYG advocates promote deescalation and avoidance, most famously Masaad Ayoob), I think you're badly strawmanning SYG perspectives as RealManTM.
The problem with DTG isn't that retreat is Unmanly, or the various pragmatic problems where a jury second-guesses split-second decisions about ease of retreat. It's that it demands a surrender of the public sphere:
If these riots somehow create a "no rights" zone, where criminals can do as they please but honest people must either stay away or submit to illegitimate violence, then their very existence is a violation of everything we stand for as a country, and it's time to clear the streets with tanks firing canister. In that case, he and I and everyone else have been lax in our duties, because this is a war.
There is no scenario where it is okay to let the criminals run rampant, and honest people are required to let them have their way. I don't care if it reduces the death rate, because that is not a terminal value. Living in peace and freedom is, and submitting to criminals makes such a life impossible.
(Or cfe here)
Even in a perfect world, where police hammer every criminal action, a ton of 'fight' happens well below the level of criminality or what should be seen as criminality. Whoever is willing to defect can commandeer large portions of the public sphere, readily. And we do not, bluntly, live in that perfect world: no small portion of coastal cities have transparently given permission for extralegal actors to crush political positions they don't like, while ignoring (sometimes 'mandatory!') restrictions on bad actors they do like.
I've recently pointed out elsewhere that I'd expected the feds to go after Rittenhouse, and that didn't happen, and it's meaningful that it hasn't. I've been meaning to do a deeper breakdown on other contemporaneous predictions I'd made, and while the retrospective looks bad, it's not maximally bad. As another example, not seeing Biden pardon Reality Winner helps suggest that there's some goals other than hitting the other team as hard and as unabashedly as possible, as much as situations like the mockery of James Gardner might feel like it.
There's space for other stuff, here. We could have seen prosecutors not immediately respond with another dip to the same well. TraceWoodgrains highlighted, as a specific contrast to calls for unilateral disarmament, the importance and possibility of a Biden administration that would "seek Republicans out and install them into the less overtly ideological spots in his cabinet", instead of dodging court mandates to throw money at people with four-year degrees. We could have some level of insight when major court cases not bother to use any more than a handwave to distinguish a Gadsden Flag from Students for Justice in Palestine with a handwave, or maybe faculty showing a bit more of a standard today. We could have people here respond to my presentations of extremely short sentencing or outright non-prosecution of serious bad actors by showing something similar and/or encouraging such leniency with the political valence flipped, or at least discomfort where pardon of an aggressive sentencing turned into a spurious bribery investigation, rather than demanding increasingly impossible statistical analysis. We could have a FACE Act prosecution on religious matters drop the hammer.
We don't, often. That sort of costly signal is costly, after all, even if not necessarily that costly.
It happens. Stranger things have happened! Just not often.
On the other direction, I'm disappointed, if not surprised, that Abbot and the Parole Board provided very little explanation or justification highlighting the specific evidence they believe justified this pardon. This isn't an analysis; it's an order. I don't think progressives would care whether Abbot was motivated specifically by scattershot of thirty different tiny details from the court record, or some bad process by the judge in question, or a sudden appreciation for limits on jury research, or whatever.
But it would at least open the possibility of debating a matter on the merits or the facts. Instead, rather than risk depending on factual claims that might be wrong, there's nothing here but frame control and leaving space for others to fight over frames.
gattsuru
7d ago
?He's set for a jury trial in 2025!? Nice problem to have if you're out on bail, but what the ever-living fuck.
There looks to have been some process stuff -- what looks like a pretty questionable police-tossed-my-evidence back and forth that took a several months, and a judge recusing because his brother-in-law turned out to be involved in analyzing the case -- but that's still nearly New York City levels of delay.
gattsuru
7d ago
Below someone said that because Foster had his gun angled down, but could have pointed it directly at Perry and fired in an instant that Perry was correct to have felt threatened. But we have video of Rittenhouse wandering around gun pointed low where he also could have brought it up and fired at any of the people around him.
I'll point to Cornered Cat for a summary that's focused on a not-lawyers-not-legal-advise, but the tripod of ability-opportunity-jeopardy is common to a much broader ethos among Red Tribers. Someone being physically able to harm you can't be a threat on its own, or everyone from a police officer to a car driver to a stick holder is cause for justifiable self-defense. Someone who says they'll hurt you can't be a threat on its own, or a trash-talking Call of Duty player would be justifiable self-defense. It's the combination of both that make for justifiable self-defense.
I think the situation for Perry is a lot more unclear, not least of all because of the low quality of all available video. But having people beating on your car doors and windows is a lot closer on jeopardy than a rando giving out bandaids (as, importantly, was Perry's driving!). Maybe not enough, and I'm disappointed that neither Abbot nor the parole board seem interested in explaining the evidence they found so compelling. But enough that it seems to be a big missing factor in a lot of the discussions and comparisons.
((That said, in turn, Rittenhouse is an obscenely good shoot for reasons that have been covered elsewhere; he set a standard that is wildly above the minimum for lawful self-defense.))
gattsuru
7d ago
Where how unique it is might matter for California's statutory right of publicity, the state's common-law right is far more expansive. I'll point to White v. Samsung, where this was close enough to trigger California's common-law right of publicity.
((Look at the decision itself for even more expansive stuff: "Here's Johnny" alone was apparently enough for the 6th Circuit to find infringement of right of publicity.))
It's an absolute mess of a standard, and celebrities have marinated in it so long that it's water to them.
gattsuru
8d ago
I'll caveat that this is a little more complicated than the quick summary -- you can find some Catholics being very skeptical and treating the accusers as heretics into the height of the early modern witch trials, and there's a controversial claim of an English witch-execution as early as the 900s. It's not clear how much the earlier Church was free of witch-hunting among the laity because they didn't believe in it (or were told to not believe in it), and how much because the records weren't made to start with.
gattsuru
8d ago
A good part of the philosophical problem with right of publicity is that it has very little impact outside of the commercial protections, and even the commercial protections are only protective to the point where the broader public knows you. It's very much a cut out to protect celebrities and the famous, not defend the little people. California has an unusually broad combination of statutory and common-law protections, but it's still only something that matters to public figures worried about getting used as an advertising or product campaign.
((Other 'moral rights' have similar problems: see VARA for a particularly ugly one.))
But that specific context impacts here, at least if Altman did what people are thinking he did.
gattsuru
8d ago
(But I'm a homo, so presumably I'm not the target audience, and maybe I'd be a big fan of some Josh Hartnett soundalike with an analogously please-fuck-me inflection, I dunno.)
I mean, there's definitely some male voice actor contributions that turn a piece much more memorable for me (eg, recent NakedSav+SpicyGayDog piece has a 'good puppy', a lot of LewdDev's work), enough that I avoid ASMR/audiobook/RVC stuff because I worry it'll be addictive.
But I don't really want that from a random app, and even as someone who would use (and has used) AI for adult content, I'm hoping that is has uses other than that.
gattsuru
8d ago
A lot of work goes into these things, cadences, pitch, pronunciation; once you're far enough in you can't change voices without changing a lot of other work.
Eh... I dunno.
Historically, yes, but a lot of the recent tools are amazingly good. This guy (cw: FFXIV spoilers up to 6.0, NSFW audio) is audibly AI-gen, but it's based on a character that has maybe an hour or two of voice lines, total, and while it's ElevenLabs rather than running on a home desktop, I'm pretty sure you could get similar results through RVC. Handling more varied content over longer periods would probably want more input media, but it's the work of days rather than months.
gattsuru
8d ago
The likely worst-case legal scenario is a lawsuit followed by settling out of court for a trivial amount.
Depends on what you consider trivial. TraceWoodgrains pointed to Midler v Ford in California, and it's foundational for Californian law, but the punchline is that Ford got off scot free, and the ad agency in question was hit for 400k USD. But that's because Midler was an issue of first impression at the time, limiting evidence of 'evil motive'; contrast the later Waits v Frito where Frito-Lay and its advertising company got tapped for a combined $2m USD over an ad that "broadcast in September and October 1988 on over 250 radio stations located in 61 markets nationwide" (though the advertising company had verbally offered to indeminfy Frito-Lay before running the ad). Contrast in turn White v. Samsung, where a literal robot acting as but clearly not Vanna White, which rhymes with today's problem, and ended up at 400k USD over a fiery dissent.
It's not business-ending, at least for a business OpenAI's size -- even adjusted for inflation and for how much Californian juries hate tech companies, I'd expect closer to 1m than 100m. But for all the philosophical problems with an expansive right of publicity, it's not toothless.
gattsuru
8d ago
Paxton's condescension at the opposition probably isn't great for the longer-term stability of the nation, but in terms of direct impact, I'm more worried about everybody else. If the future were dominated by the most authoritarian political movements in the country stomping each other's faces, well, it'd be bad, but I can't say it would be bad because of the poor innocent jerks.
It's the people in the crossfire. There's always some fuzzy edges where maybe the immigration enforcement is rough-handed to discourage illegal immigrants, or maybe the LGBT restrictions are breaking privacy For The Children, or what have you, but there's also times where people are pretty obviously hammering a matter to drum up attention, or even just because they'd be expensive in human and financial and political capital to defend, and Paxton hasn't hesitated.
Paxton's far from unique in this, and I'm not sure he's even in the top ten. If politics were the proverbial game of chicken, we've long since gone from simply throwing the steering wheel out the window, to shooting a hired driver and cutting the brake lines. But Garland's been bad enough, and we don't need two in a row.
gattsuru
8d ago
I'd caution that the NRA and its members are technically the 'victim' in the current New York lawsuit, but that didn't stop James from threatening the entire organization's mandate, digging through and almost-certainly leaking a ton of internal records, and pretty much crippling both the legal and political expenditures for one election already and probably a second. Tots coincidentally, no insurance provider in the state is willing to work with the organization, a ton of competent personnel have fled the ship or started planning competitors with all the inefficiencies and lost time that demands, so on. We won't know the full reckoning for a bit (June?), but the possibility that the org ends up under a hostile state's conservatorship is absolutely still in the cards.
Tides doesn't face that threat, but it's not because the state can't fuck over a badly operated donor funnel; it's because Republicans don't have the infrastructure to make that push.
gattsuru
8d ago
In some ways, that difference can make it a better metaphor, especially for conversations in the 1990s and early-00s. Questions like whether you can treat sexual minorities with additional caution because of an infectious disease (or even protect them from themselves, as defenders of the Cuban concentration camps sanatorios argue even today), or ethnicities with suspicion because a co-religionist drove a plane into a building are still relevant, even if they're not the central case. Rogue killing someone with a casual touch, or Cyclops blowing up a city block with a blink, are exaggerations, but there are answers to these questions that also answer all the closer ones.
I'm a fan of bringing up trans stuff and gun stuff... well, partly because it makes both sides very uncomfortable, but also because the question of whether a dick gun makes a rapist murderer drives a lot of disagreement. Not all, especially outside of the TERF border, but a decent amount. And one reasonable response is that ability alone does not make for a deadly act: it takes either decision or negligence.
It's just that this ended up not being where the broader progressive movement actually went. There had always been a fraction insistent that prejudgement was fine for even things far smaller than leveling an skyscraper, it was just being pointed the wrong direction, and they won. Once you've decided that the possibility was enough, you're pretty quickly going to find yourself just haggling over the price. At the risk of pointing to metafictional example:
Huntington's disease was a hereditary degenerative disease with cognitive and psychiatric symptoms, one of which was psychosis. Huntington's was seen in perhaps one in eight thousand people, and psychosis was seen in perhaps one in ten of those. If a randomly selected human of Superman's apparent age were to obtain Superman's pwoers, there would be in a one in eight thousand chance that they would both have Huntington's disease and the symptoms of psychosis, the result of which would probably be casualties that would dwarf the Great War by a large margin...
When these probabilities were multiplied together, the final very rough estimate was that Superman had a one in ten chance of bringing about a global scale human catastrophe of some kind in the next thirty years. Even if the odds had been one in a hundred, Lex would have taken a similarly extreme course of action.
It mostly didn't work, but settling with Defense Distributed (and giving a not-trivial amount of cash in the settlement offer) is the sort of lawfare I'd expect from a coldblooded conservative, if small-scale by the standards of that sort of cy pres-like lawfare. And then there's the obvious guesswho stuff that didn't work entirely.
I agree that a Count of Monte Cristo-style planned revenge isn't really Trump's strong point, though.
EDIT: that said, I do think it's the sort of thing Paxton would a) have the temperament and skills for, and b) absolutely do it for both political ends and to make an impulsive boss happy.
There were also a number of 'official' recordings that just disappeared, too, along with other disappearing physical evidence like the famous front door. With the noteworthy exception of Kahoe after Ruby Ridge, it's less that bringing any enforcement against FBI or DoJ destruction of evidence was tried and found hard, and more than trying them was found undesirable and left untried.
In theory, that's the next step: the House had already asked Morens twice about his personal e-mails, producing 2k with a 'voluntary' letter in November 2023, and then getting this dump of 30k pages after a subpoena mid-April 2024, and they had credible reason to believe he was sandbagging them.
But the House investigations are not criminal investigations, nor are they the FBI or DoJ. The theoretically-broad subpoena powers are limited on the enforcement side, and there's little if any executive branch support here. With a few notable exceptions all on one side of the political aisle, the threat of a contempt of congress charge is toothless unless the issuing subpoena bends over backwards about following all rules, and unlikely even then.
And one of those rules are 18 USC 2702, which generally prohibits ISPs from disclosing stored data. There are a few exceptions -- LEO have wide cutouts in 18 USC 2517, 18 USC 2511(2)(a)ii lets FISA and the attorney general do whatever they want -- and some that apply outside of a warrant, but nothing relevant here. This is also why, even though using a personal e-mail for government business makes the entire personal e-mail subject to FOIA review in civil courts (though see caveat about "under agency control", since Fauci retired), it's almost always necessary to motion for a party to the case to disclose them, rather than the ISP or e-mail service.
So, uh, mostly because they don't want to be able to.
More options
Context Copy link