It seems clear to me that the author was experiencing an unmedicated psychotic disorder and gallantly owning the preposterous outlook he had at the time. So, not bragging, just plainly stating the sort of bigger than life delusions that come with the territory.
> It reaffirmed a delusion that had previously gripped my mind and begun my psychosis; I must be dead. This place, purgatory. I prayed for a quest or that my retribution was almost over, so that I could go back to land of the living.
...
> The old lady’s came marching in, and fed us pasta and bread and cake and cookies. There was Coca Cola. I ate because I had no choice, but suspicious of the sugar they forced on us. We slept on green mats. Most of the folks, drenched theirs in Industrial Clorox. I thought they had it wrong. Embrace the filth. Do you trust the chemicals?
...
> Some of the people around me drenched theirs in mountains of maple syrup. Is sugar the enemy?
...
> My paranoia of the place had dissipated slightly as the idea of having my own room felt pretty nice.
...
> I had refused to steal, convinced I was in purgatory and doing the “right thing” was the only way to pass the test. But standing there, the rules dissolved.
This article is dripping with psychosis. It's a story about a failing health care system.
Reading some of these comments, I'm starting to understand how the system could be the way it is, and just how far we have left yet to go.
I see how it could be an actual psychosis, but I read it as full of rhetorical questions / observations / figurative speech. For example "is sugar the enemy?" could easily be a commentary on having to choose between unhealthy calories and having not enough when offered free meal. Paranoia and purgatory is phrasing often used by people for more fancy descriptions rather than literal.
Maybe if you miss that sentence, the rest can fly under the radar. Assuming one also ignores the commonly reported correlation between homelessness and mental illness. But try and have those two elements front of mind, and re-read the article: it should be abundantly clear that this is in no way a healthy person down on his luck--this is a story of someone going through a mental break.
What on earth would lead you to conclude that unbanked households don’t use online services? I can’t imagine any possible set of starting assumptions that would lead there, short of fairly cartoonish assumptions about the demographics the FDIC pointed out at that link.
Even within the unbanked households, the FDIC link points out that 1/3 use online non-bank services instead. And independently of that, it makes sense that even cash households might interface with online commercial activity: pick up gig work through DoorDash or UberEats or whatever; get paid out through a neighborhood informal-cash-service operator (multiservicio, hawala, guy who informally cashes out undocumented drivers). Or through opening a Venmo or CashApp account instead of a bank account.
That leads to a slightly stronger form of the claim: that those 5.6 million are likely to have undergone KYC/AML through other, non-bank financial providers…
But even then, why should a bank account be connected to whether or not you’re an adult in society’s eyes?
> and those 5.6 million probably aren't accessing sites that require age verification.
Why would you presume that?
> Not every solution needs to work for 100% of people.
A solution that censors large amounts of speech and culture from millions of people is clearly either insufficient or, if it is deemed sufficient, authoritarian.
I did not read anywhere that this solution can only be used if it's the ONLY solution. Did you?
How is the statement "not every solution needs to work for 100% of the people" controversial? People are different, with different circumstances and ideally there are a variety of solutions to cover all of them
Any incremental advance is better than nothing where our rights are getting eroded faster than we can contact the ACLU to start investigating whether or not we have a case. The American Right have figured out that they can DDOS the legal system with all kinds of bullshit laws that they know won't stick, but it will take everyone 10x the time and effort that they spent spewing it out.
We can't back and wait for the perfect solution that covers all corner cases and makes everyone happy and has the perfect UX. We have to fight now while we still have something to fight for.
If the system is that I have to prove my id or age for averag network connections, then the system has already failed me. The only system I am behind is a flag that some devices can send if enabled that lets the receiving party know the user is underage. Completely optional (controllable by device owner/guardian) but if received, that party must behave in a way that acknowledges that fact. It is not a perfect system, but it retains the freedoms and anonymity of the user.
My attempt at _a_ solution isn't _THE_ solution; but it seems like there's legitimately something around leveraging existing KYC infra that could get a solid 98 out of 100 - and can realistically be implemented in a realistic timeframe.
If I'm AYLO and have been cut off from 1/3 of the U.S. for the last 18 months, I'm contacting every lawyer, cryptographer, and engineer I can get my hands on to try and get _anything_ out of this concept or ones like it.
I for one enjoyed the article and found it coherent from top to bottom. The author is a historian, which means they are likely serious about justifying their claims. That might explain the thorough sourcing you're lamenting.
I didn’t find the justification serious. Adding quotes does not actually make your paper well sourced: this is exactly the reason why I think it reminds me of high school writing.
btw, there is a recent interview with an Asani dev focusing on GPUs, worth a listen for those interested in linux on apple silicon. The reverse engineering effort required to pin down the GPU hardware was one of the main topics.
just an idea, but what if the appended audio clip was reversed to ensure continuity in the waveform? That is, if >< is the splice point and CLIP is the audio clip, then the idea would be to construct CLIP><PILC.
This is exactly what we do today! It seems to work better the more you extend it, but extending it too much introduces other side effects (e.g. the avatar will start to open its mouth, as if it were preparing to talk).
Hmm, maybe adding white noise would work. -- OK, that's quite enough unsolicited suggestions from me up in the peanut gallery. Nice job on the website, it's impressive, thank you for not requiring a sign up.
All for suggestions! We've tried white noise as well, but it only works on plain talking samples (not music, for example). My guess is that the most robust solution will come from updating how it's trained.
HeyGen (and our V1 model) literally uses the user on-boarding video in the final output. See here for a demonstration of this (https://toinfinityai.github.io/v2-launch-page/#comparisons). We are not talking about that in this thread. We are trying to solve a quirk of our Diffusion Transformer model (V2 model).
Our V2 model is trained on specific durations of audio (2s, 5s, 10s, etc) as input. So, if give the model a 7s audio clip during inference, it will generate lower quality videos than at 5s or 10s. So, instead, we buffer the audio to the nearest training bucket (10s in this case). We have tried buffering it with a zero array, white noise and just concatenating the input audio (inverted) to the end. The drawback is that the last frame (the one at 7s) has a higher likelihood to fail. We need to solve this.
And, no shade on HeyGen. It's literally what we did before. And their videos look hyper realistic, which is great for B2B content. The drawback is you are always constrained to the hand motions and environment of the on-boarding video, which is more limiting for entertainment content.
