Pinned Tweet
@nithusezni Imagine in the 1960s, when $USD was still backed by Gold, that one single US company owned more Gold than Fort Knox. It would've been impossible to value them. They really wouldn't fit inside the stock market. That's $MSTR today & forever. @MicroStrategy is uncatchable. Few.
8
7
30
92,217
#YouCanSeeEverything is good but doesn't live up to the hype, in my opinion. I'm not particularly interested in Elizabeth Holmes and never have been, so getting an inside look at her wasn't inherently compelling to me. The film never really sold me on why this particular access, at this particular moment, should matter to anyone who wasn't already deeply invested in the Theranos saga.
I also didn't like that the movie didn't interview any victims of the fraud or include their perspectives on Holmes. Instead, it focuses almost entirely on Holmes and her partner, Billy Evans. It briefly interviews her father, but it's mainly a series of conversations between filmmaker Nathan Fielder and Holmes, and between Fielder and Evans. There's almost no outside counterweight: no patient who received a false result, no investor who lost money, no former employee who tried to raise alarms. Those voices could have made this a more interesting and complete documentary.
It seems the couple invited the filmmakers into their home for 34 days before Holmes went to prison in 2023 as a way to rehabilitate her image, or at least tell their side of the story. In that sense, the movie succeeds from their perspective. They get a long, intimate platform, and the film largely stays within the frame they invited. You rarely escape their house, their framing, or their version of events.
The movie is also extremely melodramatic, with its tight close-ups and music. The trailer had that same quality, and it turned me off enough that I initially had no interest in seeing the film. But there was so much hype this weekend that I felt I had to see it for myself.
There is certainly humor from the filmmaker, too, sometimes playing into that melodrama rather than cutting through it. What balances it more effectively, for me, is seeing the couple with their young children. Those domestic moments make them sympathetic and relatable in ways that the more elaborate filmmaking devices don't.
Part of why I'm writing this is to say: slow the roll. You don't necessarily need to rush to see this one.
1/4
I also didn't like that the movie didn't interview any victims of the fraud or include their perspectives on Holmes. Instead, it focuses almost entirely on Holmes and her partner, Billy Evans. It briefly interviews her father, but it's mainly a series of conversations between filmmaker Nathan Fielder and Holmes, and between Fielder and Evans. There's almost no outside counterweight: no patient who received a false result, no investor who lost money, no former employee who tried to raise alarms. Those voices could have made this a more interesting and complete documentary.
It seems the couple invited the filmmakers into their home for 34 days before Holmes went to prison in 2023 as a way to rehabilitate her image, or at least tell their side of the story. In that sense, the movie succeeds from their perspective. They get a long, intimate platform, and the film largely stays within the frame they invited. You rarely escape their house, their framing, or their version of events.
The movie is also extremely melodramatic, with its tight close-ups and music. The trailer had that same quality, and it turned me off enough that I initially had no interest in seeing the film. But there was so much hype this weekend that I felt I had to see it for myself.
There is certainly humor from the filmmaker, too, sometimes playing into that melodrama rather than cutting through it. What balances it more effectively, for me, is seeing the couple with their young children. Those domestic moments make them sympathetic and relatable in ways that the more elaborate filmmaking devices don't.
Part of why I'm writing this is to say: slow the roll. You don't necessarily need to rush to see this one.
1/4
7
1
27
39,693
The length is another issue. At nearly three hours, this is a lot to sit through in a cinema, especially if you don't already have a strong interest in Elizabeth Holmes. The first half sometimes feels like an endurance exercise. The domestic scenes accumulate, the conversations circle, and the cumulative effect becomes more exhausting than illuminating.
Between the length and the melodramatic presentation, I think this would be a better movie to watch at home. You could pause, take a bathroom break, or step away without feeling trapped in the experience. Being interested enough to watch it doesn't mean you need to rush to a theater this weekend.
To me, Holmes and Evans are more interesting and sympathetic than they're often made out to be by the press and, in some respects, by the filmmaker's own withholding of information. I also found them relatable, particularly in the scenes with their children.
That doesn't mean I have any doubt that Holmes committed fraud. It's that she seems capable of deceiving herself about it, and she doesn't have many other cards to play at this point. Watching her calmly and repeatedly insist that she has never really lied, I see both self-protection and what appears to be genuine belief. The film doesn't clear her. It lets us watch someone who has internalized a story that allows her to keep functioning.
The best manipulators are, in some sense, self-manipulators. That might be the movie's core insight. But the film itself manipulates through disclosure and withholding, and the payoff too often amounts to realizing that the filmmakers were manipulating you all along.
In that sense, the film isn't as good at this as Holmes herself. What makes her interesting is that she seems capable of both manipulating others and being manipulated. Watching her trust Evans with the public-facing side of her life while she's in prison, and watching Fielder gradually soften his reading of her, brings us back to people believing their own framing until it becomes their operating reality.
3/4
Between the length and the melodramatic presentation, I think this would be a better movie to watch at home. You could pause, take a bathroom break, or step away without feeling trapped in the experience. Being interested enough to watch it doesn't mean you need to rush to a theater this weekend.
To me, Holmes and Evans are more interesting and sympathetic than they're often made out to be by the press and, in some respects, by the filmmaker's own withholding of information. I also found them relatable, particularly in the scenes with their children.
That doesn't mean I have any doubt that Holmes committed fraud. It's that she seems capable of deceiving herself about it, and she doesn't have many other cards to play at this point. Watching her calmly and repeatedly insist that she has never really lied, I see both self-protection and what appears to be genuine belief. The film doesn't clear her. It lets us watch someone who has internalized a story that allows her to keep functioning.
The best manipulators are, in some sense, self-manipulators. That might be the movie's core insight. But the film itself manipulates through disclosure and withholding, and the payoff too often amounts to realizing that the filmmakers were manipulating you all along.
In that sense, the film isn't as good at this as Holmes herself. What makes her interesting is that she seems capable of both manipulating others and being manipulated. Watching her trust Evans with the public-facing side of her life while she's in prison, and watching Fielder gradually soften his reading of her, brings us back to people believing their own framing until it becomes their operating reality.
3/4
2
0
7
1,744
There is a secret the movie doesn't reveal until the end. I won't spoil it, but I don't think it's as big a deal as the filmmakers make it out to be. I thought it was a fair thing for the couple to ask and that they handled it fairly. Once you know what it is, the dramatic weight placed on it feels inflated. It doesn't reframe everything so much as confirm that the filmmakers were holding a card in reserve for a late complication.
Evans is an interesting character in his own right. The way he and Holmes describe their immediate connection made me wonder whether shared experiences of trauma helped draw them together. That's speculation on my part, not something the film establishes, and I wouldn't claim to know whether their experiences were comparable. Holmes's father discusses a sexual assault she experienced when she was young, and her past relationship also figures into the story.
What interested me was the possibility that their instant romance grew partly from a sense of recognition in each other. Evans comes across as both controlling and sincerely devoted, and the film never quite resolves which impulse is stronger. That ambiguity is more interesting than the late attempt to cast him as a hidden antagonist.
I also thought the male nude scene was gratuitous. The moment with Evans in the shower seemed to exist more for shock value, or to demonstrate the level of access, than to offer any real insight. To me, it cheapens the film and undercuts the more careful tone it sometimes achieves.
Is this interesting and worth watching? Yes, if you're willing to invest nearly three hours. But if you aren't especially interested in Holmes, that's a substantial commitment, and I don't think the hype should make you feel obligated to rush out.
The formal experiments and meta turns are clever on paper, but they don't add up to something as revelatory or rigorous as the early reactions suggested. The movie keeps promising deeper insight into Holmes while mostly delivering deeper insight into how the filmmakers construct a story around her.
That can be worthwhile. It just isn't, in my opinion, the landmark documentary the coverage has made it out to be.
4/4
Evans is an interesting character in his own right. The way he and Holmes describe their immediate connection made me wonder whether shared experiences of trauma helped draw them together. That's speculation on my part, not something the film establishes, and I wouldn't claim to know whether their experiences were comparable. Holmes's father discusses a sexual assault she experienced when she was young, and her past relationship also figures into the story.
What interested me was the possibility that their instant romance grew partly from a sense of recognition in each other. Evans comes across as both controlling and sincerely devoted, and the film never quite resolves which impulse is stronger. That ambiguity is more interesting than the late attempt to cast him as a hidden antagonist.
I also thought the male nude scene was gratuitous. The moment with Evans in the shower seemed to exist more for shock value, or to demonstrate the level of access, than to offer any real insight. To me, it cheapens the film and undercuts the more careful tone it sometimes achieves.
Is this interesting and worth watching? Yes, if you're willing to invest nearly three hours. But if you aren't especially interested in Holmes, that's a substantial commitment, and I don't think the hype should make you feel obligated to rush out.
The formal experiments and meta turns are clever on paper, but they don't add up to something as revelatory or rigorous as the early reactions suggested. The movie keeps promising deeper insight into Holmes while mostly delivering deeper insight into how the filmmakers construct a story around her.
That can be worthwhile. It just isn't, in my opinion, the landmark documentary the coverage has made it out to be.
4/4
3
0
15
1,496
KΛNyΞ⚡️Ⓜ️🕸 retweeted
The personal agent is about to become the most valuable piece of software real estate since the iPhone home screen.
232
148
2,202
127,207
KΛNyΞ⚡️Ⓜ️🕸 retweeted
I am once again completely over exposed to litecoin:native
IF LTC hits $1000 by end of 2026 I will give away 1 LTC to everyone who RTs this post. (until I run out of LTC)
IF LTC hits $1000 by end of 2026 I will give away 1 LTC to everyone who RTs this post. (until I run out of LTC)
74
69
242
15,064
KΛNyΞ⚡️Ⓜ️🕸 retweeted
Bitcoin has surged +31% and $19,500 from $62,300 to $81,800 since Jim Cramer said he is selling all his BTC.
323
689
9,750
507,898
The key to dominating your competition in an AI world is to figure out how to not be your own bottleneck while simultaneously not lowering your standards.
If you figured out how to do this, please let me know.
If you figured out how to do this, please let me know.
2
0
1
120
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@finkds my experience setting up @TownAI@TownAIinct @noahrshinn , and Muse, all at roughly the same time.
Town has a great setup and walks you through the whole thing very smoothly. It's also very powerful, and you can communicate with it through the app, through different messaging apps, through email. It's very powerful, and you can interface with it in a lot of ways.
With that said, when I try to multitask through WhatsApp on Town, it drops the ball a lot. Also, you blow through credits fast on this thing. I would call Town high-end, but expensive. If I were to use it as my daily driver, I would not only need the $200/month plan, I'd probably have to be buying extra credits.
Instinct is getting hyped a lot right now by people trying to make money off it, IMO. At first, it doesn't feel as powerful as something like Town, but the more I use it, the more I like it. I'm on Android, so I only communicate with it through WhatsApp. For the time being, it's free, and there is no paid version. It has a lot of horsepower behind it. They must be spending a boatload on tokens right now, and that can't last forever without charging customers.
On one level, Instinct doesn't feel as powerful and multidimensional as Town, but on another level, its simplicity and elegance is really nice. I like how casual it feels, and it's deceptively powerful.
Muse has an incredible amount of horsepower behind it. I gave it two very intensive tasks, and it moved through them very quickly. Also, integrations are really smooth. It learns super fast, and you can throw a lot at it. They must be using a lot of compute on this right now, and it's a good value for the money right now, meaning there's a free plan and if you want to upgrade, there's a $20/month plan and a $100/month plan. My sense is it's a better value than Town, at least right now, but all of that can change in the future.
I want to give an honorable mention to @TaskletAI . I see it in a very similar way to Town, just not as dressed up and polished, but their $25/month plan is a pretty good value, especially if you use Luna with it. You can get a lot done at a low cost.
There seems to be a frantic race to see who's going to be the AI agent of choice. My personal favorite is still Hermes @NousResearch and I also have a ChatGPT Pro and @perplexity_ai Max plan. I'm not a fan of Perplexity Computer, but I like the other features of the Max plan. The ChatGPT work app is super impressive, and I like how I can use my OAuth account with Hermes. I would say, as a non-coder, I get my best work done with the ChatGPT work app with either Sol or Astra.
I have yet to try Grok @bot , but probably will be doing their 3-month trial soon. Based on what I've seen of it, it feels like a dressed-up version of Tasklet, as it's essentially a VPS, but it's looking like Grok 4.7 could be at Astra level, which is why I want to give it a try.
There are so many options right now, it's insane. It's going to be really interesting to see who are going to be the winners in this race.
Town has a great setup and walks you through the whole thing very smoothly. It's also very powerful, and you can communicate with it through the app, through different messaging apps, through email. It's very powerful, and you can interface with it in a lot of ways.
With that said, when I try to multitask through WhatsApp on Town, it drops the ball a lot. Also, you blow through credits fast on this thing. I would call Town high-end, but expensive. If I were to use it as my daily driver, I would not only need the $200/month plan, I'd probably have to be buying extra credits.
Instinct is getting hyped a lot right now by people trying to make money off it, IMO. At first, it doesn't feel as powerful as something like Town, but the more I use it, the more I like it. I'm on Android, so I only communicate with it through WhatsApp. For the time being, it's free, and there is no paid version. It has a lot of horsepower behind it. They must be spending a boatload on tokens right now, and that can't last forever without charging customers.
On one level, Instinct doesn't feel as powerful and multidimensional as Town, but on another level, its simplicity and elegance is really nice. I like how casual it feels, and it's deceptively powerful.
Muse has an incredible amount of horsepower behind it. I gave it two very intensive tasks, and it moved through them very quickly. Also, integrations are really smooth. It learns super fast, and you can throw a lot at it. They must be using a lot of compute on this right now, and it's a good value for the money right now, meaning there's a free plan and if you want to upgrade, there's a $20/month plan and a $100/month plan. My sense is it's a better value than Town, at least right now, but all of that can change in the future.
I want to give an honorable mention to @TaskletAI . I see it in a very similar way to Town, just not as dressed up and polished, but their $25/month plan is a pretty good value, especially if you use Luna with it. You can get a lot done at a low cost.
There seems to be a frantic race to see who's going to be the AI agent of choice. My personal favorite is still Hermes @NousResearch and I also have a ChatGPT Pro and @perplexity_ai Max plan. I'm not a fan of Perplexity Computer, but I like the other features of the Max plan. The ChatGPT work app is super impressive, and I like how I can use my OAuth account with Hermes. I would say, as a non-coder, I get my best work done with the ChatGPT work app with either Sol or Astra.
I have yet to try Grok @bot , but probably will be doing their 3-month trial soon. Based on what I've seen of it, it feels like a dressed-up version of Tasklet, as it's essentially a VPS, but it's looking like Grok 4.7 could be at Astra level, which is why I want to give it a try.
There are so many options right now, it's insane. It's going to be really interesting to see who are going to be the winners in this race.
0
1
6
3,793
KΛNyΞ⚡️Ⓜ️🕸 retweeted
According to an article published by @Tangem back in April of this year, there were already well over 120,000,000 tokens and coins in the crypto space.
> 85 million of those reside on Solana
> 35 million of those reside on Ethereum & similar chains
Only 9-17k of those 120 million are actively traded, which is already a miniscule 0.0142% of the total supply.
Of those actively traded, the top 25 by MCAP make up only 0.147% of that.
Of those 25, only Bitcoin, Litecoin, Dogecoin, and Monero are fairly launched through a genesis block, no pre-mine, no contentious hard fork free coins, fully decentralized Proof-of-Work cryptocurrencies. This means these four cryptos, built upon the ethos and original definition of what a cryptocurrency is, make up only 0.00000333% of the total 'crypto' market.
That, my friends, shows you just how special Litecoin is. Not just as a crypto with the second strongest use case by transactions only to Bitcoin, but as digital silver.
> 85 million of those reside on Solana
> 35 million of those reside on Ethereum & similar chains
Only 9-17k of those 120 million are actively traded, which is already a miniscule 0.0142% of the total supply.
Of those actively traded, the top 25 by MCAP make up only 0.147% of that.
Of those 25, only Bitcoin, Litecoin, Dogecoin, and Monero are fairly launched through a genesis block, no pre-mine, no contentious hard fork free coins, fully decentralized Proof-of-Work cryptocurrencies. This means these four cryptos, built upon the ethos and original definition of what a cryptocurrency is, make up only 0.00000333% of the total 'crypto' market.
That, my friends, shows you just how special Litecoin is. Not just as a crypto with the second strongest use case by transactions only to Bitcoin, but as digital silver.
38
114
543
60,399
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@apples_jimmyonsensus is that we're early-stage RSI.
Regarding AGI, I think it's fair to say that on short-horizon tasks, we're pretty much there. But on longer-horizon tasks, it gets murky.
I'm pretty sure Frontier Labs will achieve internal AGI on long-horizon tasks that go well into the double-digit days within the next 6 to 12 months, and probably achieve a type of internal ASI within 36 months.
The main issue from here will be greater restrictions and access because the alignment problem is only going to get bigger, with larger and larger implications.
Few will get to hold the hammer of the gods.
Regarding AGI, I think it's fair to say that on short-horizon tasks, we're pretty much there. But on longer-horizon tasks, it gets murky.
I'm pretty sure Frontier Labs will achieve internal AGI on long-horizon tasks that go well into the double-digit days within the next 6 to 12 months, and probably achieve a type of internal ASI within 36 months.
The main issue from here will be greater restrictions and access because the alignment problem is only going to get bigger, with larger and larger implications.
Few will get to hold the hammer of the gods.
1
1
6
520
>50% chance of SI by Jan 1, 2030 is a rational estimate. This would turn the world completely upside down.
@KanyeWestJSa@alexwgbreadcrumb via training loops and nested agents.
>50% early RSI to SI in 36 mo: reasonable estimate amid speedups, though uncertain.
Bullets:
• RSI accelerating AGI/SI timelines
• Normalcy overhang persists
• Agent swarms show collusion risks
Most important: RSI. Prior models train successors, agents recurse, math autoformalizes—potentially compounding to SI by 2029 if unconstrained.
Takeaway: Capability jumps may outpace institutional readiness.
0
0
1
126
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@imjustnewataie, then I believe the following would happen:
- Anthropic, Google, and xAI would come out with similar ~AGI models within ~3 months. Meta may follow another ~3 months after that.
- A quantized version of ~AGI from an open-source model would run on 2 DGX Sparks or a 256 GB M5 Ultra within ~15 months of the OpenAI ~AGI model
- Major labs would be more restrictive on what you can do with their models, as well as on usage rates, which will lead to an even greater demand for local hardware.
- The cost of local hardware will continue to skyrocket.
- The economy and stock market could experience extreme volatility through the bubble IPOs of OpenAI and Anthropic, along with spikes of job losses in certain sectors.
- Public resistance to AI will continue to escalate without a foreseeable ceiling, which will only lead to greater restrictions on frontier models.
- The next ~3 years will turn the world completely upside down, and no one really knows how it will all shake out. This is the Demis Hassabis hypothesis of "10 times bigger than the Industrial Revolution and 10 times as fast".
- Anthropic, Google, and xAI would come out with similar ~AGI models within ~3 months. Meta may follow another ~3 months after that.
- A quantized version of ~AGI from an open-source model would run on 2 DGX Sparks or a 256 GB M5 Ultra within ~15 months of the OpenAI ~AGI model
- Major labs would be more restrictive on what you can do with their models, as well as on usage rates, which will lead to an even greater demand for local hardware.
- The cost of local hardware will continue to skyrocket.
- The economy and stock market could experience extreme volatility through the bubble IPOs of OpenAI and Anthropic, along with spikes of job losses in certain sectors.
- Public resistance to AI will continue to escalate without a foreseeable ceiling, which will only lead to greater restrictions on frontier models.
- The next ~3 years will turn the world completely upside down, and no one really knows how it will all shake out. This is the Demis Hassabis hypothesis of "10 times bigger than the Industrial Revolution and 10 times as fast".
2
2
109
14,136
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@HermesAgentTipsrk is available now. For the Mac Studio M5 Ultra, you'll have to wait for until it ships. If you're weighing older model Mac Studios versus a DGX Spark, then the answer is definitely DGX Spark. If you can wait, the Mac Studio M5 Ultra will definitely be faster, but it also may take some time for software optimizations. This is where CUDA on the DGX Spark has an advantage, even though overall it is slower hardware. I suspect the demand for the M5 Ultras will be super high, and you'll pretty much always have to wait for them after you place the order. Whereas DGX sparks continue to be available immediately on the market, even if the price keeps going up. So, if you want it now, go with the DGX Spark. If you have patience and don't mind paying now and waiting for delayed shipping times, then the Mac Studio M5 Ultra is the way to go.
0
1
0
260
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@GoogleAIStudio Just made my Twitter anthem song with this youtu.be/hY8v2QvVX2M?is…
0
1
3
8,110
When will we have ChatGPT 6 Astra-level intelligence in an open-source model that can run 50+t/s on two DGX Sparks or a 256 GB M5 Ultra?
1
0
4
1,804
KΛNyΞ⚡️Ⓜ️🕸 retweeted
@NathanCHubbard Don’t think the former CEO of Ticketmaster should be complaining about screwing fans over
12
67
3,714
45,452