Perspective as a service from a raging generalist
72 stories
·
0 followers

Fair bosses: The true “Empire State of Mind”

1 Share

Perhaps the template for the tough manager dates back to the early days of the industrial revolution. The first factory workers had spent their lives working either as artisans or farm workers, where they had some control over either the hours or the pace at which they worked. Factory owners had invested a lot of money in machinery and they wanted to get the maximum return. As a result, they insisted on strict discipline. Indeed, many factory owners preferred to hire women, or children, who were easier to control. 

The harsh rule of managers (and their agent, the factory foreman) explains why workers decided to form trade unions so they could fight to get better pay and working conditions. One industry that was plagued by industrial disputes was car manufacturing, and the result was often poor quality cars produced at high prices. Consumers had to put up with the situation because the industry was dominated by a few producers, such as General Motors, Ford, and Chrysler in the US or British Leyland in the UK. But in the 1970s, strong competition emerged in the form of the Japanese producers which initially focused on smaller, cheaper vehicles. The pioneer was Toyota, which had developed a “lean manufacturing system” designed to reduce waste and eliminate faults. A key element was a concept called jidoka, which means “automation with a human touch.” Any workers could stop the production line to prevent a fault. This required management to trust the workers to make responsible decisions, and implied a much more collaborative approach than the Western system. 

More generally, treating employees well seems to work. Alex Edmans of the London Business School conducted a study looking at those businesses ranked the “100 best companies to work for” published by Fortune magazine. The data (compiled by the Great Place to Work Institute) cover the period 1984 to 2011. After controlling for factors such as size and industry sector, the businesses that treated their staff well produced stock-market outperformance of 2.3–3.8% a year, a substantial cumulative gain. 

In his book The Art of Fairness, David Bodanis recounts a number of tales about managers who have had striking success by treating their workers with fairness and empathy. One of the most striking examples is the Empire State Building, which was built in just 13 months in the 1930s, a period which also included the dismantling of the hotel that originally occupied the site. Paul Starratt, the builder, paid his workers double the daily rate and devoted a lot of attention to safety, paying the employees in full on days when it was too windy to operate. Hot meals were delivered to the workers, even on the higher floors. The project had very low staff turnover and workers even suggested productivity improvements. 

Another example cited by Mr Bodanis was from the public sector; the 2012 London Olympics opening ceremony organised by Danny Boyle, the film director. Thousands of volunteers were needed for the ceremony but that created the risk that the details would leak. Rather than make the volunteers sign a legal non-disclosure agreement (the conventional approach), Mr Boyle trusted them to keep the secret. The volunteers lived up to their part of the deal, partly because Mr Boyle treated them in a grown-up fashion, listening to their suggestions and ensuring they were paid for their costumes. 

By and large, the services sector requires a collaborative approach between workers and managers. If the job requires any kind of creativity, or use of initiative, it is fruitless to attempt to turn the workers into automatons. An interesting example is Timpsons, the British high-street chain that specializes in key cutting and shoe repair. Sir John Timpson, the founder, believes in “upside-down management” which involves trusting the staff to serve the customers in any way they want. They are allowed to change prices and settle disputes. The main two rules are “look the part” and “put the money in the till”. 

By and large, the services sector requires a collaborative approach between workers and managers.

Of course, there will always be workers who will take advantage of employers and “swing the lead”. (This metaphor comes from the days when sailors used a piece of lead on a line to check the water depth; lazy mariners would simply swing the lead in the air rather than dip it in the ocean.) In a services-based economy, checking a worker’s productivity is less easy than it used to be; managers cannot count how many widgets they have produced or bricks they have laid. 

In some parts of the economy, it can be impossible to improve productivity. An economist called William Baumol described this problem; it takes the same number of musicians the same amount of time to play a Schubert string quartet as it did when the piece was first produced. But despite this lack of productivity improvements, musicians’ wages have risen in real terms over the centuries. That is because those who employ musicians need to offer a wage sufficient to deter them from taking a job as, say, a plumber or a teacher. In a sense, workers in low productivity sectors “free ride” on the back of improvements in more competitive industries.

This article Fair bosses: The true “Empire State of Mind” is featured on Big Think.

Read the whole story
Jakel1828
141 days ago
reply
DFW
Share this story
Delete

No Thanks

iA
1 Share

Last week, Apple got into trouble because of an ad. It taught us a lesson beyond the good and evil of advertising and public relations.

The ad showed various objects being collectively crushed by a hydraulic press: a chronometer, a record player, a trumpet, a piano, a guitar, paint, an arcade machine, LPs, notebooks, photo cameras, and other objects. At last the press opens to reveal the new iPad.

The idea: iPad compresses creativity.
But we were not impressed.

The ad triggered outrage that led to an apology. Part of the scandal was predictable, avoidable, and yawnable. But there was something unexpected about it.

Raging Against The Machines

A joke that wasn’t funny

Apple wanted to tell us that their new iPad somehow contains all those creative tools and icons of entertainment. It was an ad, a joke — but then it wasn’t funny. It tried to lie metaphorically, but it literally told the truth.

An iPad does everything and nothing well. It’s the eternal teenager of computers, a computer that doesn’t want to grow up and commit.

What we saw was Apple destroying beautiful things that have a clear creative and entertaining purpose — unlike the iPad. A trumpet makes music. An arcade box lets us play games. A notebook lets us write. They do one thing, and they do it well. And, unlike the iPad, they last.

How about playing it backward?

The ad might have worked, had they played it in reverse. Then, the message could not have been taken literally. In reverse, the ad would show creative tools magically growing out of an iPad and suggest that they mythically live inside the device.

The ad forward was the truth pretending to be a joke. Backwards it would just be another pretty lie about how this, latest version of a gadget that never made you happy, would finally, really make you happy.

The reason why we all got so upset had little to do with cinematography. Film tricks can hide but not undo the truth: that more technology won’t automatically make us happy or solve our problems.

Why exactly did we get so upset?

It doesn’t take a French deconstructionist to connect the dots between the Apple ad and the avalanche of Tik-Tok videos with people demonstrating craft. Destroying LPs, pianos, and trumpets is literally and metaphorically tone-deaf.

Watching people cooking, making music and painting is becoming more popular than another artificial beauty posing in front of a rented Lambo. In the age of Is this real? we prefer to watch regular people sing, sculpt and dance.

It’s nice to see real people playing a real guitar through an artificial screen but, it’s really great if it happens right in front of us, better even: with us. Your grandma’s risotto cooking tips have outdone Ivanka Trump’s makeup secrets.

Being reasonable

We are learning

The outrage against Apple last week though came from different places. There was the Social Media Rage, nostalgia, and the usual tech skepticism. But there was also a simple “no thanks.” We can just say “no thanks” to technology if it doesn’t make sense. New is not inevitable. New doesn’t mean need.

We are learning to say no thanks to 3,500 Dollar VR goggles, subscription-based AI pins, and orange plastic boxes that want to book hotels for us without checks and balances.

We are realizing that we don’t need to look like an electro-gadget Christmas tree to prove that we are ready for the future. We found out that we don’t need every generation of every gadget ever produced.

We can say no thanks to the absolute blockchain, the total Internet of Things, the celestial AI. We can say no thanks to anything we don’t really like or need — and eventually, we will.

Finding (better) analogies

At some point, we switched off our TVs and forgot about The Bold and the Beautiful. We threw out our fax machine. One day, our kids just put their fidget spinners away. It can be hard, but it’s easier if you have something more meaningful to do.

It’s easier to leave Twitter if you have a great book. It’s easier to press pause on Spotify if you can play the trumpet. It is easier to avoid Uber Eats if you know how to cook. It is easier to close your computer if you know how to swim a butterfly.

Switching off and getting real can make us happier. Being real also means that eventually you will also feel really sad. And that one reason why we choose phones over pianos. Here’s the good news:

Getting it

Your no thanks doesn’t need to be an absolute, total, eternal no. We can use typewriters to write letters and write a poem on our phone at the bus stop. We can say no thanks to AI phone bots but still use ChatGPT helping us to understand Aristotle’s Metaphysics in the original. We can be both critical and appreciative of new technology without categorically accepting or rejecting it.

Knowing what you want doesn’t mean that you don’t “get it”. What do we tell our kids when they first pick up a hammer, knife, or a trumpet? Don’t hurt yourself. Be careful. Don’t break it. Being careful with tools, new and old, is common sense. Welcome back to that.

The post No Thanks appeared first on iA.

Read the whole story
Jakel1828
141 days ago
reply
DFW
Share this story
Delete

Everything’s a cult now

Vox
1 Comment
Illustration of three silhouettes, each with a hole in its head. A red ribbon runs through all three holes.
Getty Images

Derek Thompson on what the end of monoculture could mean for American democracy.

Is damn near everything a cult now?

That’s a glib distillation of an interesting idea I recently encountered. The basic thesis was that the internet has shattered the possibility of a monoculture and the result of that is a highly fragmented society that feels increasingly like a loose connection of cults stacked on top of each other.

To say that everything is a cult is a bit of an overstatement, but as a general framework for understanding the world at the moment, it is helpful. The way we consume content, the way fandom works, the ways we sort ourselves into tribes and camps online, even the way lots of industries work, including the news business — it all has shades of culthood. This is easier to see if you set aside the more extreme examples of cults, like the ones that end in mass suicide or shootouts with the ATF, and instead think of cults as movements or institutions that organize themselves around the belief that the mainstream is fundamentally broken.

Understood this way, there are lots of cults, or cult-adjacent groups, and not all of them are bad. But if society keeps drifting in this direction, what will that mean for our shared democratic culture? How much fragmentation can we sustain?

To think all of this through, I invited Derek Thompson to The Gray Area. He’s a staff writer at the Atlantic, the host of the podcast Plain English, and the person who originally floated this idea about the cultification of society. Below is an excerpt of our conversation, edited for length and clarity. As always, there’s much more in the full podcast, so listen to and follow The Gray Area on Apple Podcasts, Spotify, Stitcher, or wherever you find podcasts. New episodes drop every Monday.


Sean Illing

Tell me why you think everything’s become a cult.

Derek Thompson

I’ve always been very interested in culture, which I suppose is worth defining. Culture is the way that we think about the world and the way that we influence each other’s thoughts about the world. And that can be through entertainment, it can be through religion, it can be through fashion and clothes, but it’s the memes and ideas and ideologies that not only influence our own sense of reality but other people’s sense of reality. And I’ve always been interested in how people’s sense of reality comes to be.

So you can start with the late 19th century when the concept of a national reality was first possible, at least in America. You had technologies like the telephone and the telegraph that allowed newspapers to share information and report on information that truly was national. It allowed information to travel much faster than it had ever traveled before. And so suddenly in the late 19th century, we had the possibility of a national and even international real-time shared reality. And that shared reality might have come to its fullest expression in the middle of the 20th century with the rise of television technology. You had just a handful of channels that were reaching tens of millions of people.

At the same time, you also had the rise of national newspapers and maybe the apogee of national newspapers in terms of their ability to monopolize local advertising revenue and become enormous machines for getting tens of millions of Americans to read about a shared reality. And so you move from the 19th century with the birth of this possibility of a shared reality, to the 20th century, where you really have the rise of a kind of monoculture, which was never really possible for the vast majority of human history.

What I’m interested in is the possibility that the internet has forever shattered that reality, that we are in a way going back to the pre-20th century where culture is actually just a bunch of cults stacked on top of each other, a bunch of mini local realities stacked on top of each other.

Sean Illing

How do you define a cult?

Derek Thompson

I think of a cult as a nascent movement outside the mainstream that often criticizes the mainstream and organizes itself around the idea that the mainstream is bad or broken in some way. So I suppose when I think about a cult, I’m not just thinking about a small movement with a lot of people who believe something fiercely. I’m also interested in the modern idea of cults being oriented against the mainstream. They form as a criticism of what the people in that cult understand to be the mainstream. And cults, especially when we talk about them in religion, tend to be extreme, tend to be radical, tend to have really high social costs to belonging to them.

Today, especially in the media and entertainment space, we have this really interesting popularity of new influencers or new media makers adapting as their core personality the idea that the mainstream is broken, that news is broken, that mass institutions are broken, that the elite are in some way broken and elite institutions are broken. The fragmentation of media that we’re seeing and the rise of this anti-institutional, somewhat paranoid style of understanding reality, I see these things as rising together in a way that I find very interesting.

Sean Illing

You were talking about the phone and the telegraph earlier, but the thing about newer technologies like radio and TV, for instance, is that they helped create something like a mass culture. The public was more or less watching the same movie we call reality, and for all the downsides of that, and there were many, it did have the benefit of grounding society in a shared reality. Do you think of that loss as a genuine cultural and political crisis? Or is it possible that this is just another period of technological change, not that different from earlier periods and we’ll figure it out?

Derek Thompson

I do think that in so many ways, we’re just going back to the middle of the 19th century. We’re going back to the historical norm rather than being flung into the exosphere, into some unprecedented state of popular discombobulation.

The idea that a shared reality, a shared national reality in real time, is even possible is so historically young. Just one quick aside, I was doing some reporting for the book that I’m writing right now and saw in an Eric Hobsbawm book called The Age of Revolutions that when the Bastille fell in 1789, a Canton 30 minutes away from Paris didn’t realize the French Revolution had happened for a full month. That was the speed at which information used to travel. It was the speed at which a man could ride a horse or walk next to his horse.

You need a whiz-bang technology that can somehow transmit at something like the speed of light, certainly one would hope the speed of sound, information across vast distances. You only had that with the invention of the telegram and the telephone, and then later radio.

So if you want to know where we’re going, look where we came from. In the 19th century, of course, we had lots of chaos, but we also had an American democracy for decades and decades. So it’s not obvious to me that the erosion of the monoculture or the erosion of the news mainstream is anathema to American democracy.

Sean Illing

I don’t think it’s incompatible with American democracy as such, but it might be incompatible with the model of liberal democracy we’ve become accustomed to since mid-20th century or so, which is also a historical aberration.

Derek Thompson

You might be totally be right. This is one place where the bridge goes too far for me to have a ready-made answer. I’m not exactly sure why a more riotously antagonistic and fragmented news ecosystem would be perilous to liberal democracy. It’s possible that it would be, but what’s the causal mechanism by which a wildly fragmented media leads to a backlash in liberal democracy?

Sean Illing

I guess I’m thinking about how this environment creates a collective action problem that makes dealing with the sorts of challenges we’re dealing with today almost insoluble. Martin Gurri had a useful metaphor in his 2018 book The Revolt of the Public. The way he put it was to say that for a long time we looked into the same mirror of mass culture and the internet shattered that mirror into a billion little pieces, which meant that governments could no longer dictate the stories societies were telling about themselves, which is a great thing in lots of ways, but it also produced a lot social turbulence.

Derek Thompson

I agree with the idea that we’re all looking into fewer mirrors, but it’s not obvious to me that the mirrors we were looking into were reflecting reality. They were reflecting a version of reality that left out a lot. The news of the 20th century did not report on racial justice at anything like the level of quality that we now expect reports in racial justice to do. The mirrors of the 20th century and news reports of the 20th century did not, I think, uncover all sorts of problems of governance that took years to understand. Didn’t report on the environmental degradation of industrial America in the 1930s, ’40s and ’50s. Protests had to fill the void of media that was under-representing minorities in urban America.

None of this is to accuse you personally of overlooking those problems because I’m sure you would agree with all of them, but it’s to remind all of us that when we feel nostalgia for the media environment of the 1940s and 1950s, we are feeling nostalgia for a news media ecosystem that in many ways was inferior to and even blind to the problems that we know to pay attention to today. And I do think that in many ways, the fragmentation of the media can sometimes create competition that allows us to see behind corners and understand things, root out problems that we didn’t see before.

I’m a capitalist overall, and I think that more competition in most markets is good. I just think it’s important to understand, as we do in some markets, that there can be negative externalities. A huge gaping negative externality of abundance in media is that superabundant media creates a scenario where news entrants feel like they have to be antagonistic. A news environment like that is going to create a lot of distrust, it’s going to create a lot of disharmony, it’s going to confuse a lot of people, and it will replace a world with a small number of flawed mirrors with a riotous and unthinkable number of mirrors, some of which are absolute bullshit mirrors and some of which are quite good.

Sean Illing

The problem of “distrust” is what I was getting at. I never liked the phrase “post-truth” because it implied there was a golden age in which we lived in truth. That’s bullshit. So I’m not nostalgic in that way and I’m not making the case that we understood our world better, or that society was more just, when everyone was watching the same handful of networks or reading the same handful of newspapers. I’m just saying that was a period where there was more trust in authority, in part because of this near-monopoly on information at the top. And when that near-monopoly shattered, people could see and hear more and that eroded trust in authority, trust in experts, trust in information. Is that a good thing in the long term? Probably. I don’t know. But I don’t think our institutions were equipped to manage the transition from that world to this one.

Derek Thompson

I think I agree with a lot of that. What I most want to hold down on is the idea that almost all nostalgia for a past golden age is nostalgia for a world that did not exist or a world that we would find inexcusably terrible today. If someone believes that the world of 1950s or ’60s was better in this way, then why didn’t that shared reality lead to a world where we fixed our problems faster? Why didn’t a shared reality more expeditiously reveal the injustice of Jim Crow and voting laws before the 1965 Act? Why didn’t it help us see the terrible things that we were doing to leaded gasoline and the air and the water? Why, essentially, was the world of monocultural news so flawed if monocultural news is so useful for showing the electorate what’s important in the world? That’s the question that I feel like is never answered when people start waxing nostalgic about the middle of the 20th century.

To hear the rest of the conversation, click here, and be sure to follow The Gray Area on Apple Podcasts, Spotify, Pandora, or wherever you listen to podcasts.

Read the whole story
Jakel1828
142 days ago
reply
'If someone believes that the world of 1950s or ’60s was better in this way, then why didn’t that shared reality lead to a world where we fixed our problems faster? Why didn’t a shared reality more expeditiously reveal the injustice of Jim Crow and voting laws before the 1965 Act? Why didn’t it help us see the terrible things that we were doing to leaded gasoline and the air and the water? Why, essentially, was the world of monocultural news so flawed if monocultural news is so useful for showing the electorate what’s important in the world? That’s the question that I feel like is never answered when people start waxing nostalgic about the middle of the 20th century.'
DFW
Share this story
Delete

Expectations Versus Reality

1 Comment

A few months ago, OpenAI showed off “Sora,” a product that can generate videos based on a short prompt, much like ChatGPT does for text or DALL-E does for images, and I asked myself a pretty simple question:

"...how can someone actually make something useful out of this?" and "how do I get this model to do the same thing every time without fail?" While an error in a 30-second-long clip might be something you might miss, once you see one of these strange visual hallucinations it's impossible to ignore them.  

A month later, OpenAI would debut a series of short films, including one called “Air Head," a minute-and-twenty-second-long piece about a man with a balloon for a head, one that changes sizes 23, 24, 26, 27, 29, 32, 34, 39, 41, 42, 43 and 45 seconds into the piece, at which point I stopped counting because it got boring. 

The very nature of filmmaking is taking different shots of the same thing, something that I anticipated Sora would be incapable of doing as each shot is generated fresh, as Sora itself (much like all generative AI) does not “know” anything. When one asks for a man with a yellow balloon as his head, Sora must then look over the parameters spawned during its training process and create an output, guessing what a man looks like, what a balloon looks like, what color yellow is, and so on. 

It repeats this process for each shot, with each “man with balloon as his head” character subtly (or not-so-subtly) different with each repetition, forcing users to pick the outputs that are the most consistent. Achieving a perfect like-for-like copy isn’t guaranteed, but rather, filmmakers must pick the ones that are “good enough.”

This becomes extremely problematic when you’re working in film or television, where viewers are far more likely to see when something doesn’t look right — a problem exacerbated by moving images, high-resolution footage, and big TV screens.

Yet the press credulously accepted Sora’s “stunning” videos that were “amazing and scary,” suggesting to the public that we were on the verge of some sort of AI takeover of the film industry, helping buoy Sam Altman’s attempts to “convince Hollywood that Sora won’t destroy the movie business.” These stories only serve to help Sam Altman, who desperately needs you to believe that Hollywood is scared of Sora and generative AI, because the more you talk about fear and lost jobs and the machines taking over, the less you ask a very simple question: does any of this shit actually work?

The answer, it turns out, is “not very well.” In a piece for FXGuide, Mike Seymour sat down with Shy Kids, the people behind Air Head, and revealed how Sora is, in many ways, totally useless for making films. Sora takes 10-20 minutes to generate a single 3 to 20 second shot, something that isn’t really a problem until you realize that until the shot is rendered, you really have absolutely no idea what the hell it’s going to spit out. 

Sora has no mechanism to connect one shot to another, even with “hyper-descriptive prompts,” hallucinates extra features where you haven’t asked for them, and Shy Kids were shocked by how surprised OpenAI’s researchers were when they requested the ability to use a prompt to request a particular angle in a shot, a feature that was initially unavailable. It took “hundreds of generations at 10 to 20 seconds a piece” to make a minute and 19 second long film, with “probably 300:1 in terms of the amount of source material to what ended up in the final.” 

Yet these tiny little problems all lead to one overwhelming issue: that Sora isn’t so much a tool to make movies as it is a slot machine that spits out footage that may or may not be of any use at all. Almost all of the footage in Air Head was graded, treated, stabilized and upscaled, as that 10-20 minute lead time on generations was for footage in 480p resolution, meaning that even useful footage needed significant post production work to look good enough. 

To put it as plainly as possible, every single time that Shy Kids wanted to generate a shot — even a 3-second-long one — they would give Sora a text prompt, and wait for at least ten minutes to find out if it was right, regularly accepting footage that was subprime or inaccurate, the best example being that many of Air Head’s shots are in slow motion, a hallucination that Shy Kids never asked for but had to accept because the shot was otherwise what they needed. This caused them to have to do “quite a bit of adjusting ...to keep it all from feeling like a big slowmo project,” which I’d argue it still does.

 Sora’s problem isn’t really the amount of time it takes to generate a shot, but the imprecision that comes from using a generative artificial intelligence that doesn’t really know anything. Patrick Cederberg of Shy Kids described this as each prompt “...[showing] another interpretation of that latent space,” and that Sora has only learned the aspects of what something looks like rather than actually knowing what it is, effectively coming up with a fresh take every single time while creating media for a medium that relies on consistency. And to be clear, Shy Kids was trying to do something fairly simple — a balloon attached to a human body — rather than, say, trying to have the same person or character look and move and act the same way across a contiguous series of shots.

To digress, none of this shit actually works if you’re making an actual movie. While Pixar movies may take years to render, it has supercomputers and specialized hardware and, more importantly, the ability to actually design and move characters in a 3D space. When Pixar renders a shot,  it isn’t flipping a coin to see if Sully from Monsters Inc will look the same way between one shot and another, nor is it worried that the other monsters will look weird, or different, or move differently, or if the shot will be in slow motion. While there might be some unexpected things that happen, these are things that can be solved because somebody actually created the characters' models, and there are tools that exist to manipulate them in space.

Sora’s attempts to replace filmmakers are dead on arrival because these are impractical and ineffective solutions to problems that nobody complained about other than Hollywood executives. The AI hype bubble, as I’ve noted before, is one entirely reliant on us accepting the idea of what these companies will do rather than interrogating their ability to actually do it. Sora, much like other generative AI products, suffers from an imprecision and unreliability caused by hallucinations — an unavoidable result of using mathematics to generate stuff — and massive power and compute requirements that are, at this time, prohibitively expensive for any kind of at-scale usage.

To make Sora kind of useful, OpenAI will have to find a way to dramatically increase the precision of prompts, reduce hallucinations to almost nothing, and vastly increase processing power across the board. Sora hasn’t even been launched, save to a few hand-picked companies that got to test an early version, meaning that the 10-to-20 minute wait between generations is something that’s likely to increase once more people use the product, and that’s before you consider how incredibly expensive it is to run a significantly more complex model than ChatGPT, which is already unprofitable for OpenAI to run

These are also the intractable problems that OpenAI has failed to solve. OpenAI failed to make a more efficient model for Microsoft late last year, and while GPT-5 is meant to be “materially better,” it isn’t obvious what “better” means when GPT-4 performs worse at some tasks than its predecessor. I believe Sam Altman is telling us the truth when he says that the future of AI requires an energy breakthrough, but the thing that I think he’s leaving out is that it may take an energy breakthrough (and more chips) for generative AI to approach any level of necessity, hoping that people will keep buying the hype without asking too many annoying questions, like “what does any of this stuff actually do?” or “is this useful?” or “does this actually help me?”

To be clear, Sam Altman is the single most well-connected and well-funded man in artificial intelligence, with a direct connection to Microsoft, a multi-trillion dollar tech company, and a rolodex that includes effectively every major founder of the last decade, and he still can’t get past any of these very obvious problems, partly because he isn’t technical and thus can’t really solve any of the problems himself, and partly because the problems he’s facing are burdened by the laws of mathematics and physics. 

Generative AI hallucinates because it doesn’t have a consciousness or any ability to learn or know anything. Generative AI is extremely expensive because even the simplest prompts require GPT-4 to run highly-complex mathematical equations on GPUs that cost $10,000 apiece

Even if this were cheaper, or more efficient, or required less power, it would still be a process that generates answers based on the (extremely complex) process of ingesting an increasingly dwindling amount of training data. These problems are significantly compounded when you consider the complexity, size and massive legal ramifications of training a model on videos, a problem that nobody has seen fit to push Altman about.

That’s ultimately the problem with the current AI bubble — that so much of its success requires us to tolerate and applaud half-finished tools that only sort of, kind of do the things they’re meant to do, nodding approvingly and saying “great job!” like we’re talking to a child rather than a startup with $13 billion in funding with a CEO that has the backing of fucking Microsoft

We’re all expected to do the mental and emotional labor for Sam Altman and other AI companies, saying that it’s “remarkable that they’re able to do even this” and both assume and give them the credit for some inevitable future where all of these problems are gone, despite little proof that such a thing is possible.


The Rot Economy’s growth-at-all-costs mindset has allowed the tech industry to fully disconnect from meaningful value creation, creating illogical public markets that pump stocks based on vibes and fund private companies that sound good enough without anybody asking whether what they’re planning is useful, or marketable, or, well, possible. 

Late last year, a shadowy tech firm called Rabbit would raise about $30 million across two funding rounds to make, and I quote, “a natural language-powered personalized operating systemwhich connected to something called a “Large Action Model” that Rabbit would leverage to “understand human intentions, interface, interactions, and reimagines the relationship between man and machine.” In January, it would debut the R1, a $200 box that with a little screen that responds to voice commands, and receive an obscene amount of frothy, unquestioning press for a device that claimed it could, with a simple voice command, order you both dinner and an Uber home, while also operating as a voice-based assistant like ChatGPT. 

This device is (was?) promising to take minute actions in apps that require your username and password, using only your voice. When demoing this device, at no point did CEO Jesse Lyu, let alone any of the reporters talking to him, actually demonstrate Rabbit’s “Large Action Model,” nor did any of them question the outright ridiculous idea that you’d be able to point this thing at an image and ask it to edit out a watermark, and it’s frankly shameful that I can’t find a single member of the tech press who wrote “wait, this all sounds like bullshit, right?”

I’ll note that I’m leaving out that the R1 has a weird little camera that can look at stuff and tell you what it is, and something involving having it look at a spreadsheet and make edits based on your commands. Who cares! Who gives a shit! Nobody needs this crap! 

In fact, the very idea of a Large Action Model is deeply, deeply suspect. This company that raised $30 million less than a year ago has suddenly popped up out of nowhere to show us that it’s created an artificial intelligence model that can control your apps as if by magic, yet there don’t seem to be any demos of this “Large Action Model” beyond Lyu’s extremely awkward attempt to have his R1 generate a prompt in Midjourney. Putting aside flowery language about how the Large Action model “emphasizes [Rabbit’s] commitment to better understand actions, specifically human intentions,” nobody seemed to be asking the very, very simple question — how and where would it actually be manipulating these apps? How would they be controlled? How was nobody asking about the fact that Rabbit claimed to have an entire new kind of artificial intelligence model?

Or perhaps the truth was a little simpler: there wasn’t much artificial intelligence happening at all.

An alleged leak of the Rabbit R1’s source code discovered by student Michael Golden heavily suggests — and I’ve confirmed this with others too — that Rabbit’s “Large Action Model” is actually repurposing Playwright, a piece of software that allows you to automate actions in a virtual machine (basically a simulated operating system that exists in the cloud) using scripts, allowing you to duplicate common actions like choosing a song in Spotify. The biggest issue for Golden, who works in responsible AI development, is that the Rabbit appears to create a fresh virtual machine where your login details are entered, and said login details are managed by a third party (in this case, a company called Privacy Piiano, and that was not a typo) rather than industry-standard tools like OAuth, which Golden says are “very simple” to integrate. 

In essence, you are trusting a third party that nobody has ever heard of to handle your logins, and trusting them every time you interact with the R1. This would, of course, mean that if Rabbit’s servers were ever hacked, whoever was on them would have access to whatever instances were live at the time, along with complete, unfettered access to your apps. As Joseph Thacker, security researcher and Principal AI Engineer at AppOmni pointed out, Rabbit users face three main threats: 

  • Rabbit’s staff could potentially misuse the user credentials stored on the virtual machines. This model creates an opportunity for a malicious insider to cause serious harm.
  • If one of those machines is infected with malware, such as a keylogger, an external attacker could easily steal user credentials. 
  • If Rabbit fails to properly secure user credentials, or an attacker gains access to every virtual machine, they could compromise the accounts of every Rabbit customer. 

I’d also like to raise a few concerns of my own, after consulting with some more technical people:

  • First, I don’t see how it’s possible for Rabbit to hash and salt user credentials. This bit needs explanation. When you create an account on a website, the website doesn’t (or shouldn’t) store the password in plaintext, but rather in a form where it’s impossible for even the website to see the password. The password is transformed from the original form (like “hunter2”) to a long string of characters and it can’t (at least, not easily or cheaply) be reversed into the original form. Since Rabbit is acting as a third-party, and doesn’t use a standard like Oauth or SAML, it needs to keep user credentials in their original form. This is just insanely dangerous. 
  • Thacker hints at this point in his blog, but it’s worth saying directly here: If the user doesn’t use two-factor authentication on their Facebook or Google account, or whatever account they use to sign-in to other websites, an attacker could use that account to compromise other accounts — even those the user didn’t share with Rabbit. Admittedly, this concern is purely theoretical at this point, with the device only working with a handful of apps (namely Spotify, Doordash, Uber, and Midjourney), but something to consider for the long-term.  
  • Even if the user has two-factor authentication enabled, this isn’t a silver bullet. Many websites, to make life easier for users, have something called adaptive multi-factor authentication (MFA). In practice, this means that the website will only ask for a code if it appears the user is signing-in from a new device or location, or if there’s something else manifestly different that sparks suspicion. Adaptive MFA works well, but like any AI system, it isn’t perfect.

But more fundamentally, Rabbit demands an exceptional level of trust from users, who must share the credentials to their apps, and yet, the company — and its founder, Jesse Lyu — isn’t exactly forthcoming with users. Rabbit didn’t start life as a left-field consumer tech company, but rather a nebulous NFT/metaverse project called GAMA that raised millions and made bold promises (which included sending an actual satellite into space), but was swiftly abandoned as soon as the market soured on NFTs and shifted to AI. This origin story has been largely memory-holed by Rabbit and Lyu, and only really came to light by examining remnants of the GAMA project, which included Clubhouse recordings they failed to delete

Every ounce of scrutiny piled on Rabbit in recent weeks has suggested that this product is badly-designed, or at the very least, not particularly sophisticated in its design. The hardware itself runs AOSP — the Android Open Source Project, which is essentially the core Android operating system, minus Google’s proprietary code and apps. In fairness, this isn’t too unusual, as it’s one of the more common IoT platforms, alongside various Linux distributions, FreeRTOS and its various derivations, and, to a lesser extent, the Windows IOT family. 

But here’s where it gets entertaining. The Rabbit software is an Android app, and some hackers have managed to extract the code from the device and port it to other devices, including a Google Pixel 6a. This was accomplished, in part, by the company’s failure to actually introduce effective measures that would limit the software to Rabbit-made hardware

Even more amusingly, some hardware hackers have managed to install LineageOS — a third-party version of Android designed to offer a lightweight, bloatware-free experience — onto the device, turning it into an actual functional Android device — save for, that is, the sluggish six-year-old MediaTek Helio P35 CPU, which was originally intended for the cheapest of cheap smartphones.

Putting aside the remarkable security concerns, the “Large Action Model” also looks like barely-functioning dogshit. Rabbit’s live stream of the launch, which mysteriously isn’t available on YouTube or any other platform, features the worst tech demo I’ve seen in my life, where (at around 36:17) CEO Jesse Lyu attempts to order McDonald’s. Nearly a minute later, absolutely nothing has happened, which Lyu blames on DoorDash’s interface, before saying that the R1 “gave him a random restaurant,” which he dismisses. He tries again, and a full minute later, the R1 manages to show him a selection of things that he can order from McDonald’s, which he then hits “order” on, at which point nothing else seems to happen. After hesitating, Lyu refreshes the computer in front of him, which shows that he has successfully added a meal to his cart on DoorDash, and then successfully completes the transaction, causing the audience to half-heartedly cheer. 

He then adds that “ordering food is no big deal from your phone,” a task that the Rabbit R1 took several minutes to execute.

As an aside: Last week, Twitter user Xyz3va — a software developer and infosec worker, according to her bio — posted footage showing several unauthorized apps running on the virtual machines that power Rabbit’s Large Action Model. These included Minecraft and (of course) Doom, the classic 1993 FPS that’s been ported to lawnmowers, thermostats, and (kinda) pregnancy tests.

Xyz3va accomplished this because of a fundamental flaw in the design of Rabbit — namely, the use of VNC (remote computing) to access the device’s cloud-based back-end services. While what she did is probably technically illegal, it’s deeply concerning that it proved so incredibly easy, thanks to the genuinely bizarre design choices made by Lyu and team. This can’t be overstated: most cloud-connected apps and hardware devices don’t use a remote desktop to communicate, but rather APIs, which offer a way to interface directly with other apps and services directly.

Xyz3va also raised further doubts about the “AI” bit of the Large Action Model, with the backend (which is a stock-standard Ubuntu virtual machine, with a few apps installed) using the Playwright testing framework to interact with the likes of Uber, Spotify, and so on.

Playwright is a useful tool for web developers. If you’re testing a website, you want a way to check that pages and functionality works across multiple browsers and rendering engines. Instead of performing those tests manually, Playwright lets you write scripts that perform those tests, with actions like clicking a button or pressing text represented as lines of JavaScript.

Like I said, useful. But it isn’t AI. And using Playwright like this also presents obvious challenges for Rabbit. If, say, Uber changes the layout of its website, or changes the flow of certain actions (like booking a cab, or changing the pickup location), it would effectively break that entire part of the service. The Rabbit team would have to effectively re-write that code. 

Other features on the R1 — voice notes that generate summaries in text (a product offered by at least three different companies), live translation (that still requires you to select a language) and a dorky little camera that can tell you what you’re looking at (a feature in multiple different apps for years that doesn’t even appear to work properly) — are interesting yet at no point make this device particularly impressive. 

Worse still, it lacks very, very basic features that you’d expect in a device from a decade ago — timers, reminders, calendars, photos, access to your emails, things that one would assume were obvious for an “AI assistant,” unless of course the people making the assistant aren’t really building things for real people, because they don’t talk to them. And despite having a touchscreen, the R1 doesn’t let you use it, requiring you to use its frustrating-to-use scroll wheel, which YouTuber Marques Brownlee theorizes is to stop you comparing the R1 to the smartphone you already have.

And putting aside all of the minutiae — what is the point of this device? Practically, why do I need this? Say that every feature on the R1 worked perfectly — and pretend that it somehow didn’t hallucinate, it always knew exactly what it was looking at, and it could order you DoorDash with your voice. What person actually needs this? What function does this device have other than those already offered by smartphones released five years ago? Who is the user and how do they use this? How does this fit into a real person’s life, and what function does it serve? 

Even then, how could anybody buy the premise of the Large Action Model? Why did anybody humor the idea that a $200 plastic box — one without a subscription, one from an unknown company — would be able to magically control all your apps with voice? I’m not going to go through every single article that was written about the R1 before its announcement, but multiple outlets that are now crushing this device credulously accepted the “Large Action Model” as a done deal, which I would argue let ten thousand people buy this thing without really being given the truth. 

Practically-speaking, how would this have worked, other than having a literal script (or person) clicking around the app to do the things you wanted it to? And how would that possibly ever be reliable or consistent, let alone faster or more efficient than you doing it on your phone? Wait, you’re telling me that people can train the Rabbit to do these tasks using “train mode,” but the train mode isn’t even available? And that that training will train that Large Action Model that definitely for sure exists and isn’t just a Large Language Model talking to a Playwright script? What’re you talking about?

YouTuber Marques Brownlee, who recently eviscerated Humane’s $700 AI Pin, declaring it “the worst product he’s ever reviewed,” raised a very important point about the R1 in his review: that we, as consumers, keep being shipped devices like cars, games and gadgets that are half-finished with a vague promise of “it’ll come at some point,” with AI products “at the apex” of the trend. Marques asks a simple question: what do reviewers do? Are you supposed to give these products the benefit of the doubt?

The answer is no, Marques. The answer has always been no. 

Companies that deliver features late should not get credit for features they are yet to deliver, and companies like Rabbit that engage in the language of science fiction to paper over their lazily-developed half-solutions should get buried for doing so. And when these companies promise far fetched ideas without actually demonstrating them — especially when said ideas sound extremely hard to execute — they should be met with poignant skepticism and then damning criticism when they fail. 

Why does the media — or worse, the consumer — have to sit there and take it as they’re shipped half-finished shit? Would Rabbit accept if instead of $200 I gave them $125, with the vague promise that my $75 would be “on the way in the future”? And why can’t the media call these companies out before they deliver? While I’m not calling for outright skepticism of literally every idea, the R1 was always making a ridiculous promise — that it’d just magically use your apps with your voice. This was always going to suck! This was obviously, clearly, transparently a ridiculous thing to say you can do! 

If you can get past that, even then Rabbit is just a shitty, awful business. The $200 Rabbit R1 is made by Teenage Engineering, a famed design house that made the Playdate console, which CEO Cabel Sasser referred to as having “surprisingly slim” margins. Assuming that the R1’s hardware makes Rabbit $50 per sale (and that is being very, very generous), Rabbit still has to pay for access to the Large Language Model that powers the R1 (which allegedly uses Perplexity AI for searches, though I can’t find anyone talking about using that on their device), as well as the cloud compute costs for running the so-called Large Action Model’s cloud-based scripts. 

While Rabbit has tens of millions of dollars of venture funding, Rabbit’s business is about as functional as the R1, and I would bet money that it finds a way to try to charge its 100,000 or so customers some sort of subscription fee to stay afloat, just after the single most conspicuous tech reviewer called it “barely reviewable.”

As an aside, one of the biggest critiques of the R1 is that even if all the pieces did work, it really would make more sense as a standalone app. Funnily enough, this is exactly what the R1 is — an Android app shoved into a bright orange box.


It’s time — it’s been time for a while — to accept that artificial intelligence isn’t the digital panacea that will automate our entire lives, and that reality may be far more mediocre than what Sam Altman and his ilk have been selling. 

Brownlee remarks in his review that he really wants a “super personalized AI assistant that can do whatever a human assistant could,” and I really want to be clear how far we are off from something like that, because something that can automate your calendar, your emails, your music and your travel in a way that you can actually rely on requires operating system-level access and processing of information that may never exist. 

Large Language Models are — and always will be — incapable of reasoning or consciousness, and while Sam Altman is desperate for you to believe that AI agents will soon run your apps, there’s little evidence that these agents will ever exist in a form that resembles a “supersmart assistant.” How’re they gonna pull that off? Training it on hours of people clicking buttons? There are a million different edge cases in every single consumer app, and reporters kvetching about the fear of these agents taking people’s jobs should sit and consider whether these things are actually possible.

Every single story about artificial intelligence seems to begin with “well, AI can’t do much yet, but…” and then lean into sloppy science fiction that masks that these “autonomous agents” are effectively chatbots that can answer questions based on a document, tech that’s existed for a decade or more. Agents that can automate repetitive tasks have been around for far longer.  UiPath, a profitable Robotics Process Automation company, was founded in 2005, and makes hundreds of millions of dollars selling access to software for Windows to automate minute processes. UiPath does a great job obfuscating what it is despite publishing a multitude of different case studies, because the reality is far more drab — it automates data entry tasks like putting stuff into Salesforce, and each automation requires precise instructions, coding expertise and more than likely some sort of service contract from UiPath. One would think that if this was technologically possible, UiPath, or Blue Prism, or any number of RPA companies would’ve done so.

Except it just isn’t possible, especially using a supposed Large Action Model that trains on off-the-shelf software testing scripts, if that’s even how the LAM works. AI agents are error-prone, and the likelihood of failure in an automated task like moving a calendar invite or ordering an Uber is exponentially higher and less tolerable in a consumer setting. Generative AI models are not going to suddenly start controlling your apps, or your calendar, or your email, or run distinct parts of your life, and those even suggesting they can are engaging in journalistic malpractice.

And when you really sit and think about it, the current AI hype cycle has done very little for society other than helping quasi-con-artists make millions of dollars. What has ChatGPT done? What has it changed, and for whom has that change been positive? What, exactly, does a regular person get out of Google’s Gemini and its ability to recognise that you drew a duck on a piece of paper — something that Google has already admitted was edited to remove how long it took to work? 

Every day, regular people are forced to deal with cold, hard reality. If you fail to complete a task at work you’ll run the risk of being fired. If you fail to pay a credit card bill, your credit score drops and you can get sent to collections. If you fail to pay your mortgage, you run the risk of losing your house. Normal human beings in their personal and professional lives constantly deal with the fact that failure is very rarely an option, and that failing to deliver on a promise is met with harsh consequences, oftentimes at the hands of richer, more powerful people, all while the media tells us that people aren’t working hard enough at their jobs.

Yet rich men with millions or billions or trillions of dollars behind them are continually applauded for delivering semi-useful bullshit, backed by a media that can’t seem to offer the same charity and patience to people that actually work at a real job. Companies like Google, OpenAI and Rabbit should be treated with suspicion when they make flowery promises, and derision when they continually fail to do anything useful.

The tech industry needs to start building things for real people again. Solve real problems. Help regular consumers, not the shadowy morasse of Small-To-Medium Enterprises. If artificial intelligence is truly the future, build something that makes somebody who doesn’t give a shit about technology sit up and listen instead of creating increasingly-more-complex party tricks in the hopes that it’ll show Wall Street you’re still growing.


God damn I hate this shit. I am sorry to rant, I am sorry to be foul-tempered and foul-mouthed. But every day real people are working more to earn fewer dollars that don’t go as far, and the tech industry’s response is to create obtuse bullshit that doesn’t help anybody. Why does the Humane Pin exist? Who is that for? How did Imran Chaudhri and Bethany Bongiorno ship out such an unbelievably useless piece of shit? Is this a case of abject negligence or abject opportunism?

People aren’t tired of looking at their phones — they’re tired of their phones being stuffed full of notifications and spam texts and their inboxes being filled with marketing emails from companies that forced them to hand over their email address to read an article or get an offer. Nobody wants a second device to “use their phones less” and “be more present,” they want the shit they already bought to work better. 

And most people that I talk to are deeply tired of the tech industry. Every single person I know is aware of how bad Google is, how bad YouTube is, how many scam texts they receive, and how much spam is in their email. They’re aware that Facebook sucks and Instagram sucks and Uber is full of little buttons to sell you more services and that the tech industry, on some level, works against them to get rich. 

What we consider “the tech industry” seems entirely disconnected from creating technology, rewarding career con artists like Sam Altman or management consultants like Sundar Pichai, nakedly corrupt businessmen that have no interest in actual technological progress. Tech is no longer a playground of “crazy ones” or experiments or cool new gadgets, but a merry-go-round of EBITDA analyses and considerations of how we can package up half-finished or actively harmful things and sell it, sometimes without even making a profit. Venture capital investment is at a five-year-low because an alarming amount of VCs don’t seem to understand technology or business, and seem entirely married to hype cycles. 

I admit I was kind of excited to see the Rabbit R1 launch, if only because it harkened back to the mid-2010s when it still felt like weird little freaks were making weird little apps. Yet Rabbit’s R1 is just as cynical a con as that perpetuated by big tech — a series of naked lies papered over with cutesy design, Jobsian presentations, and the vague promise that maybe the thing you paid for will work one day.  

People like Jesse Lyu are the problem. There are founders that actually make things — functional apps, real businesses — that could use the funding and the attention that the Rabbit R1 received, yet Lyu, someone that has spent years making outlandish promises he fails to deliver on, is the one that got $30 million, Lyu is the one that got all of this attention, and Lyu is the one that will now see his company publicly burn as a result of launching half-finished shit, damaging public trust in the tech industry in the process. 

I recognize it’s dramatic to call someone a liar, but it’s very important that I do so. People like Jesse Lyu continue to succeed when they aren’t held accountable. I will gladly retract my accusation if Lyu can actually deliver a product that does everything he says, one that functions properly and reliably and securely. As recently as a month ago, Lyu told TechCrunch’s Devin Coldewey that his “neural symbolic network” can control “800 apps.” Where are the apps, Jesse? Why are you talking about Tesla not training its autonomous driving software on stop signs, did you not read about how it’s missed those in the past? Why are you telling Inverse that you “don’t want any trash designs running on your OS” when it appears that the product itself is trash? Jesse, did you use your goddamn product? You just made $20 million off of these things. Gird your goddamn loins and be honest about your failures.

And Jesse - by all means, if you’re feeling hurt or frustrated by my comments, you have an open invite to come on my podcast Better Offline. 

This is the rot in the tech industry — cults of personality enabling cults of personality to pretend that they’re doing great things as they crap out the next great contributions to our nation’s landfills. By all means be excited about what they might be building, celebrate their actual victories, but exile anyone who dares to try and tell you sell you their half-baked dreams.

Read the whole story
Jakel1828
142 days ago
reply
'The tech industry needs to start building things for real people again. Solve real problems. Help regular consumers, not the shadowy morasse of Small-To-Medium Enterprises. If artificial intelligence is truly the future, build something that makes somebody who doesn’t give a shit about technology sit up and listen instead of creating increasingly-more-complex party tricks in the hopes that it’ll show Wall Street you’re still growing.'
DFW
Share this story
Delete

Practicality vs. Possibility

1 Share

iPod and iPhone sold on practicality.

But iPad, Apple Watch, and Apple Vision Pro were all sold on possibility.

People found iPod and iPhone worth the price without an App Store. For the other form factors they’ve produced since then, Apple has asked customers to imagine what each of these could be used for, not necessarily what they already can do out of the box.

That’s not a question for customers. That’s a question best answered by Apple. I would like to see them answering that question during development of the product rather than asking everyone else to fill in the gaps they left. iPad may have been an instant hit, but Apple Watch took years to find its footing. Apple Vision Pro has a long way to go, I suspect.

I don’t think the product was rushed, but it is missing the practical component. No one feels like they need it.

Read the whole story
Jakel1828
143 days ago
reply
DFW
Share this story
Delete

The Longer the Better Club

1 Share




The Exxon aquisition of Pioneer is done and now its time to make a plan.


Exxon paid $64 large for this big block of acreage in Saudi America and it needs to have a good outing this time out; better, for instance, than the XTO mess and hopefully better than the BOPCO/Bass acquistion in Eddy County where it can't seem to meet production guidance and is now in it's 3rd request for additional time. Its current CEO promised 1 MM BOEPD in 2018, by 2023. Now its 2 MM BOEPD by 2027


This was 2023's promises from the same fella to do better...



And right on schedule, a year later, from the same guy, here is 2024's promises..



Anyway, lets take a look at how 2023 went for Exxon in the Delaware Basin, given this fine-tuned, new frac techology and longer lateral stuff...

Exxon's 2023 IP's were lower in the Delaware than previous years, its 10-12 month decline rates much steeper, and EUR likely much lower.


This is a TRRC GIS map of Exxon's new postion in Saudi America. Click the thingy. If you go to the TRRC GIS website you can see all of Midland Basin on a bigger scale and try and count 3 mile laterals, requiring more that 3 sections, drilled to date. You'll be surprised how few you can find.



A reminder, please, four mile laterals require 3 1/2 to 4 sections of minerals, not two, or three. A little over 83% of all HZ production in this urban-like neighborhood in the Midland Basin comes from just three benches, the Lower Spraberry and Wolfcamp A and B. The correlative interval covering these three benches is only about 1,300 feet thick.


Most of these long laterals are allocation wells, or production sharing wells, a scheme to force pool minerals in Texas where there is no force pooling. Some minerals are good with that, others are not but can't do anything about it. We are on a mission from God in Texas, to drain America first and there is NO statutory law, no Statewide regulatory rule than cannot be broken.


Longer laterals are all about making a quick buck, not maximizing recovery of oil in place. Exxon profits are falling and try as they may, it just can't get to its lofty production guidance. Even, believe it or not, on a BOE basis. With the Pioneer acquisition and its shit in Eddy County, New Mexico, Exxon is set to become the largest natural gas producer in North America.


You you can make a lot of money making stuff up, some $40MM a year, in fact.


Everybody thinks longer is better and everybody wants to get in the club. Even with bolt on acquisitions its still very crowded out there and this fad is not going to last long.


Next year's promise from Spring will be five mile laterals on 660 foot BHL spacing with new, nuclear fusion, single-stage frac'ing techniques, guaranteed to rattle the entire Permian Basin with 6.0 earthquakes. The world needs American tight oil, by God. Lets rock !




Read the whole story
Jakel1828
143 days ago
reply
DFW
Share this story
Delete
Next Page of Stories