billiam

Please. The International Seabed Authority has been almost entirely captured by the same mining interests that are trying to perversely argue that only by destroying the last effective carbon sink on the planet can can we save the planet. After all, the only reason we had to kill almost every whale in the world in the nineteenth century was because we needed all that oil for lamps forever, since there was no prospect of a better lighting technology ever being developed. We will manage to find or replace these metals without hoovering the deep ocean.

Don't know any particulars but the real problem here is the misaligned incentives that come from the rather incestuous relationship between founders, investors, and their friends in most companies steroidally pumped-up by VCs. What about the customers and what they want? VCs typically over-rotate on founders they talk to and obsession with comparable companies. Reading between the lines here the real story of a fire sale to Airtable (in some trouble of its own) might have to do with burnout but also machinations with another portfolio company from a lead investor.

As the emails make clear, Musk reveals that his real goal is to use OpenAI to accelerate full self driving of Tesla Model 3 and other models. He keeps on putting up Google as a boogeyman who will swamp them, but he provides no real evidence of spending level or progress toward AGI, he just bloviates. I am totally suspicious of Altman in particular, but Musk is just the worst.

The article is actually a great illustration of how far ahead of machine learning humans remain in their ability to collaborate and make intuitive connections (the AI contribution was minimal). Which LLM is going to say, hey this pattern looks like the birds out my window, or the problem I worked on years ago and never got anywhere, or I must send a competing LLM a preprint of my paper?

Cheer up. It's not real. Generative AI is going to force us to confront what makes us human and the real world real, and learn to love it all over again. Sure, a lot of people will be lost in digital realms. Some might even like it. But I think that many will embrace the messy, imperfect, poignant realm we live in.

I find creepy things in all the videos, despite their breathtaking quality at first glance. Whether it is the way the dog walks out into space or the clawlike hand of the woman in Tokyo, they are still uncanny valley to me. I'm not going to watch a movie made this way, even if it costs my $0.15 instead of $15.00. But I got tired of Avatar after watching it for 20 minutes. Maybe all the artificial abundance and intellectual laziness the generative AI world will make us realize how precious and beautiful the real world is. For my kids' sake, I hope so.

For all you founders and YC aspirants who build companies to get rich as quickly as possible: think of what this man achieved and the lives that have been changed by creating something built to last. By not extracting maximum labor from employees and maximum cash for yourselves but by recognizing you matter less than you think to your company's success and your employees matter more. I buy Bob's Red Mills flour a few times a year and probably will for the rest of my life. TextPayMe, Loopt, Infogami, Memamp, Simmery- do we even remember what those first YC companies tried to do? Think about making things people will remember in a decade, or even 4.

Nilay's video review is just fantastic-one of the best consumer technology reviews in written or multmedia form ever. Walt would be proud.

Go ahead and read this book if you are fascinated by the psychology and character and life of Sam Bankman-Fried. I can't imagine a more boring subject--the man-baby is one of the most depressing public figures I have come across in my time in tech. I've seen his selfishness, self-absorption, and privilege many many times among startup founders. What Lewis is actually selling here and what he is always peddling in his books is a breezy, readable way to understand complex subjects for a readership who can't be bothered to learn the details, and crypto is a perfect candidate. His act is wearing thin, and The Blind Side exposed just how lazy his approach appears to be. He can pretend this book isn't about how clueless people like him can be taken in by a recognizable conman and sociopath, but that is what this book is about.

I’ve tried them a couple of times—sorry to say they’re trash. The middle tube can be under cooked while the frilly part is falling off and over cooked. Design schools and product management should teach this as a textbook example of design thinking taken way too far. Ask and ask carefully: what problem does this product solve? Maybe there is no single pasta shape that is perfect for all sauces because there doesn’t need to be.

A great organization would like some help with some outstanding technical challenges that will help further their open source research mission. Some cool asks for image analysis and signals processing as well as managing large datasets and semantic search.

I envision a future where maglev trains whisk you to ice fishing.

It's gonna be super useful to train on Uranus data to help with understanding exoplanets, since many will be like Uranus.

The organization that gave the world Rust, Web Assembly, and countless other useful technologies is in a probable death spiral, and most of the blame goes to Mitchell Baker. If Mozilla was not primarily driven by greed and laziness, smart people that still exist in around Firefox would have made the difficult technical decisions needed to continue to ship a useful product that protects user privacy while still being easy to support in a world dominated by Chrome and Safari.

Russia has put a lot of effort into developing the ability to cut and disrupt as well as intercept undersea cables, so the increased dependence on Svalbard for satellite traffic is a control point all the interested countries should be looking at. They have already demonstrated that they are willing to hold world commerce at ransom, so even as they recede as a naval power they are upping their asymmetric capabilities on the world's oceans.

Entertaining stuff, even if it makes you seem a bit more crazy (but you know, a good kind of crazy). Rock on.

I dunno, one of the slam dunks was the spectrum cliff showing vegetation. What if the energy production system on Planet X wasn't based on Mg-bearing chlorophyll? And plants have only been around for 1/7th to 1/4 of the planet's history, so it's a crap shoot any way you slice it.

Reduce 50% of all hospital deaths even moderately and I'd say you should get a Nobel prize.

Huh? ERCOT might like the idea in the short run, but as the article points out the opacity of the mining operations actually makes forecasting demand and supply harder over time, since unlike 1 million houses in a heat wave the mines can just switch on (or off) instantly. States need to regulate these operations the way they do generators or batter storage failciities, which they dwarf in size.

If you grew up and still believe the only thing a daily comic strip is for is making you laugh, I truly feel sorry for you.

It's a strange formulation. I think the point is that significant inputs of energy or chemical potential are needed to make chemical systems behave like biological systems in terms of the permanent, replicable changes that Darwin called natural selection. That's how you get disequilibrium in such usually reversible mechanisms, which the authors imply is necessary for the origin of life. I may be missing a larger point, but these energy inputs are assumed in most origin of life theories, from Urey's lightning in a swamp to more recent clay replication ideas.

Was there. Baker has constructed (and deconstructed) Mozilla to provide her with money and power. Full stop. The Google search deal is an easy way to keep Brussels off Google's back. Any normal Board would have fired her many years ago for non-performance. Brendan was popular with most engineers but a small but vocal minority had problems with a single political contribution he made years before and his somewhat cavalier way of explaining it. IMHO he was unsuited as a Mozilla CEO but he has made Brave what he wanted Mozilla to be, so happy ending.

Guess who owns the companies that bought the governments that subsidize coal production for the last century and a half? These hand wavy arguments about scale ignore the fact that the superrich and their ability to capture governance and policy are the whole problem. Put a real carbon tax on their consumption and they will not "reflect on their choices" but could become more supportive of energy policies that don't doom their customers and themselves.

It's true it's not a monolith, but the loudest voices here have been against all union,s not just public sector ones. But do you hear that today? Silence. It's the sound of former well-paid engineers and other tech workers of the techbro-libertarian bent realizing that the worker rights they've spent years decrying actually matter. Could this be because they finally realize that despite their notion that they are in the ownership class with their options and investments, their jobs have more in common with a UPS driver than with management and the superrich who own the companies where they work?

He tries a little too hard to make gradual global changes more tangible (what does 930 Hiroshimas even mean?), but in general he does a good job of stating the obvious for us lunkheads. The developed world is nearly at civil war in large part about the tiny number of refugees we have today---we will be completely unable to deal with climate refugees in the tens or hundreds of millions, and it won't be wet bulb temps that finish our society off, it will be us doing it to each other.

For those of you gathering Altman pronouncements to train your AltmanGPT model that spits out 100% disingenuous and self-serving quotage, here is what he said that is sooooo sensitive:

OpenAI’s plans according to Sam Altman

Last week I had the privilege to sit down with Sam Altman and 20 other developers to discuss OpenAI’s APIs and their product plans. Sam was remarkably open. The discussion touched on practical developer issues as well as bigger-picture questions related to OpenAI’s mission and the societal impact of AI. Here are the key takeaways:

1. OpenAI is heavily GPU limited at present

A common theme that came up throughout the discussion was that currently OpenAI is extremely GPU-limited and this is delaying a lot of their short-term plans. The biggest customer complaint was about the reliability and speed of the API. Sam acknowledged their concern and explained that most of the issue was a result of GPU shortages.

The longer 32k context can’t yet be rolled out to more people. OpenAI haven’t overcome the O(n^2) scaling of attention and so whilst it seemed plausible they would have 100k - 1M token context windows soon (this year) anything bigger would require a research breakthrough.

The finetuning API is also currently bottlenecked by GPU availability. They don’t yet use efficient finetuning methods like Adapters or LoRa and so finetuning is very compute-intensive to run and manage. Better support for finetuning will come in the future. They may even host a marketplace of community contributed models.

Dedicated capacity offering is limited by GPU availability. OpenAI also offers dedicated capacity, which provides customers with a private copy of the model. To access this service, customers must be willing to commit to a $100k spend upfront. 2. OpenAI’s near-term roadmap

Sam shared what he saw as OpenAI’s provisional near-term roadmap for the API.

2023:

    Cheaper and faster GPT-4 — This is their top priority. In general, OpenAI’s aim is to drive “the cost of intelligence” down as far as possible and so they will work hard to continue to reduce the cost of the APIs over time.
    Longer context windows — Context windows as high as 1 million tokens are plausible in the near future.
    Finetuning API — The finetuning API will be extended to the latest models but the exact form for this will be shaped by what developers indicate they really want.
    A stateful API — When you call the chat API today, you have to repeatedly pass through the same conversation history and pay for the same tokens again and again. In the future there will be a version of the API that remembers the conversation history.

 2024:

    Multimodality — This was demoed as part of the GPT-4 release but can’t be extended to everyone until after more GPUs come online.

3. Plugins “don’t have PMF” and are probably not coming to the API anytime soon

A lot of developers are interested in getting access to ChatGPT plugins via the API but Sam said he didn’t think they’d be released any time soon. The usage of plugins, other than browsing, suggests that they don’t have PMF yet. He suggested that a lot of people thought they wanted their apps to be inside ChatGPT but what they really wanted was ChatGPT in their apps. 4. OpenAI will avoid competing with their customers — other than with ChatGPT

Quite a few developers said they were nervous about building with the OpenAI APIs when OpenAI might end up releasing products that are competitive to them. Sam said that OpenAI would not release more products beyond ChatGPT. He said there was a history of great platform companies having a killer app and that ChatGPT would allow them to make the APIs better by being customers of their own product. The vision for ChatGPT is to be a super smart assistant for work but there will be a lot of other GPT use-cases that OpenAI won’t touch. 5. Regulation is needed but so is open source

While Sam is calling for regulation of future models, he didn’t think existing models were dangerous and thought it would be a big mistake to regulate or ban them. He reiterated his belief in the importance of open source and said that OpenAI was considering open-sourcing GPT-3. Part of the reason they hadn’t open-sourced yet was that he was skeptical of how many individuals and companies would have the capability to host and serve large LLMs. 6. The scaling laws still hold

Recently many articles have claimed that “the age of giant AI Models is already over”. This wasn’t an accurate representation of what was meant.

OpenAI’s internal data suggests the scaling laws for model performance continue to hold and making models larger will continue to yield performance. The rate of scaling can’t be maintained because OpenAI had made models millions of times bigger in just a few years and doing that going forward won’t be sustainable. That doesn’t mean that OpenAI won't continue to try to make the models bigger, it just means they will likely double or triple in size each year rather than increasing by many orders of magnitude.

The fact that scaling continues to work has significant implications for the timelines of AGI development. The scaling hypothesis is the idea that we may have most of the pieces in place needed to build AGI and that most of the remaining work will be taking existing methods and scaling them up to larger models and bigger datasets. If the era of scaling was over then we should probably expect AGI to be much further away. The fact the scaling laws continue to hold is strongly suggestive of shorter timelines.

That's what God thought too. Didn't stop Her.

I miss those old magazines, what a medium for transmitting ideas. I love Tesla but he really had some cra-cra ideas, particularly about Earth and oceans as conductors.

Unfortunately the CZU Fire did a number on the densest parts of the Skyline to Sea Trail. It is still beautiful but won't look like Jurassic Park in our lifetime.

>And what do you do for families whose homes were destroyed? Of the nearly 1500 structures destroyed, only a handful of building permits have been issued to rebuild.

You can't let them rebuild in place. Your points are spot on (a new survey shows 1/3 of Californians live in the urban/wildland interface where wildfires will destroy most dwellings) and it requires new zoning, new fire legislation on fire insurance, and of course the ability of people to live in higher density through infill and careful planning. The cost of housing in the Bay Area is a closely related problem, but the bottom line is that we have to let fire move through the coastal redwoods every few decades, and that means we cannot build tinderbox houses right next to those groves.