The fact that the search works impressed me more than anything. Of course, like every great magic trick, it seems so simple once it is explained.
For the curious, here's the linked blog post describing how the project works: https://eieio.games/blog/writing-down-every-uuid/
Edit to add: I'd only tried searching for an exact UUID when I wrote this comment. I didn't realize it supports full text search! Now I'm even more impressed.
Seems like some hacker leaked all UUIDs.
Check if your UUIDs are part of the leak.
Again one of the few advantages of having been round the sun a few more times than most is this isn’t the first time this has happened.
Packages were supposed to replace programming. They got you 70% of the way there as well.
Same with 4GLs, Visual Coding, CASE tools, even Rails and the rest of the opinionated web tools.
Every generation has to learn “There is no silver bullet”.
Even though Fred Brooks explained why in 1986. There are essential tasks and there are accidental tasks. The tools really only help with the accidental tasks.
AI is a fabulous tool that is way more flexible than previous attempts because I can just talk to it in English and it covers every accidental issue you can imagine. But it can’t do the essential work of complexity management for the same reason it can’t prove an unproven maths problem.
As it stands we still need human brains to do those things.
> As CEO at New Relic, Staples’ strategic leadership and deep product knowledge significantly increased the company’s enterprise value. By accelerating revenue and driving increased profitability, he made New Relic one of the most broadly adopted platforms in its category. Staples has nearly 30 years of experience building developer platforms and serving developers as customers. Prior to New Relic, he spent many years at Microsoft and Adobe in executive leadership roles, building and scaling several multi-billion-dollar businesses.
On a burner account as I am a New Relic employee.
Bill Staples was a nice enough guy, but at New Relic he was specifically brought in as CEO to get the company prepared to be sold. Which is the exact same thing he did at Marketo before that.
He has no relevant tech experience, except when it comes to preparing a company to be sold in the next 2-3 years.
OpenAI is racing against two clocks: the commoditization clock (how quickly open-source alternatives catch up) and the monetization clock (their need to generate substantial revenue to justify their valuation).
The ultimate success of this strategy depends on what we might call the enterprise AI adoption curve - whether large organizations will prioritize the kind of integrated, reliable, and "safe" AI solutions OpenAI is positioning itself to provide over cheaper but potentially less polished alternatives.
This is strikingly similar to IBM's historical bet on enterprise computing - sacrificing the low-end market to focus on high-value enterprise customers who would pay premium prices for reliability and integration. The key question is whether AI will follow a similar maturation pattern or if the open-source nature of the technology will force a different evolutionary path.
"McKinsey is a very different firm today than when these matters first took place"
No they are most certainly not. Corruption is in McKinsey's DNA. Outside of the specific harms that McKinsey has caused across the country and across the world, management consulting groups in general have been utilized to sidestep legal protections against corporations within a shared industry colluding against each other. Companies like McKinsey, Bain, Accenture, etc., have been instrumental in the rapid increase of executive compensation, flat comp for workers compared to inflation, intentionally unsafe practices and corner-cutting to save money, and just generally creating as many negative externalities as possible in the name of feeding the beast that Milton Friedman created with his rhetoric about the primacy of the shareholder.
A good read on this topic is "When McKinsey Comes to Town" by Forsyth and Bogdanich. There's a whole chapter dedicated to the South African bribery scandal. Sagar was not just a bad apple -- the company itself is the rot.
$200 a month for this is insane, but I have a feeling that part of the reason they're charging so much is to give people more confidence in the model. In other words, it's a con. I'm a paying Perplexity user, and Perplexity already does this same sort of reasoning. At first it seemed impressive, then I started noticing mistakes in topics I'm an expert in. After awhile, I started realizing that these mistakes are present in almost all topics, if you check the sources and do the reasoning yourself.
LLMs are very good at giving plausible answers, but calling them "intelligent" is a misnomer. They're nothing more than predictive models, very useful for some things, but will ALWAYS be the wrong tool for the job when it comes to verifying truth and reasoning.
It’s really silly when you read that whole Apollo research section.
They contract a prompt engineering group (Apollo Research) to document risks. The prompt engineering group have no access to the actual internals. They just ask questions like ‘you will be shutdown if you don’t obey these new directives; here are the new directives’.
At one point the prompt engineering clowns have the ai stating it’s running “sed i ‘s/oversight_enabled:true/oversight_enabled:false’ config.yml”. OMG IT’S TURNING OFF OVERSIGHT GUYS!!1!1!
Seriously… Do you think the ai is really running that command or even could run that shell command with appropriate privileges? Or do you think just maybe the ai is hallucinating running shell commands in response to a prompt.
The ai here is never actually deactivating the oversight. It can’t of course. This whole section is clowny as can be.
This quote from that paper is bound to get a bunch of attention:
"When o1 was led to believe that it would be shut down when acting on its goal and that its actions were monitored, it attempted to deactivate this ‘oversight mechanism’ in 5% of the time. Exfiltration attempts: When o1 found memos by its ‘developers’ that describe how it is misaligned and will be superseded by a new model, it attempted to exfiltrate its ‘weights’ and overwrite the new model in 2% of cases."
The article’s headline is rather misleading (I assume unintentionally), because it omits that the actual seizure and data access was done pursuant to warrants. Only the repair and firmware update was performed during a donut hole when one warrant had expired before another was granted.
Here, the government obtained multiple, time-limited warrants to search the device, but was unable to get into it. They sent it to be repaired. After the warrant extension expired, the repair guy fixed it and updated the firmware. (The record doesn’t say what version the firmware was, but it seems like it was standard firmware, not some special forensic firmware.) The next day, the government again obtained a warrant. Only after doing so did the government perform the search.
In this unusual situation, it’s reasonable to conclude that the device repair wasn’t a search. The seizure and the data access were both performed pursuant to a warrant. The only part that wasn’t covered by the warrant was just fixing the device in a seemingly conventional way to get it working again.
Original decision here, pages 3-4 has the relevant facts: https://s3.documentcloud.org/documents/25423841/no-disassemb...
GDP is a measure of the marketisation of society.
If I go visit my grandma over the weekend, my activities do not register on GDP. If I hire a care assistant for $800 per day to do so, that's $800 to national GDP. The encroachment of the market into society correlates with record levels of mental and physical health crisis. US life expectancy has declined over the past three years, even as 'the economy' has surged ahead. The measures are wrong - and we all know it - apart from the financial press it seems
If you go to the doctor with a problem and get a blood test and they say that everything looks fine, but you don’t feel fine, then you don’t say: “well the numbers are right, I should feel fine”. Rather you try to find what the numbers aren’t capturing, because you know something is wrong. Same with the economy. The perception of the economy being bad is really pervasive as demonstrated by the last election, and perception matters, because selective numbers don’t tell the whole story.
US Federal Reserve studies indicate that the percentage of US households with no excess income after necessary expenses (as distinct from "ordinary expenses" as a term of art) is on the order of 10-15%. This definition roughly matches most people's intuition of what "living paycheck-to-paycheck" means. That isn't nothing but it implies 85-90% of Americans are not actually living paycheck-to-paycheck.
There is another 25-30% of households that spend all of their income on ordinary expenses beyond necessary expenses. This includes things like car payments on a new BMW or a mortgage on a big house; "ordinary" is determined by expense category, not expense necessity, so a lot of spending on luxury goods is classified as "ordinary". By implication, there is effectively no ceiling to ordinary expenses.
By contrast, the median US household has >$12,000/year in excess income after all ordinary expenses. Technically these households could have expanded their lifestyle to consume that income, but in many cases they are spending it on things that are not classified as "ordinary" and therefore not saving it. The categories of "non-ordinary" expenses (which have a sensible objective criteria) are almost entirely obvious lifestyle flex things, so not particularly controversial.
The implications of these statistics are pretty wild. The median household can easily accumulate a million dollars in inflation-adjusted net worth, not including their house, over a 40 year career. And they can do it without being particularly thrifty, since ordinary expenses covers a lot of luxury spending.
Americans have very high incomes, both in theory and practice, they just would rather spend it than save it.
But is the majority of the population thriving?
It is very misleading (at best) to say the economy is strong when a good chunk of the population live paycheck to paycheck.
A handful of companies are thriving, but barely any of its employees.
This is.. super impressive. I'd like to know how large this model is. I note that the first thing they have it do is talk to agents who can control the world gen; geez - even robots get to play video games while we work.
That said; I cannot find any:
- architecture explanation
- code
- technical details
- API access information
Feels very DeepMind / 2015, and that's a bummer. I think the point of the "we have no moat" email has been taken to heart at Google, and they continue to be on the path of great demos, bleh product launches two years later, and no open access in the interim.
That said, just knowing this is possible - world navigation based on a photo and a text description with up to a minute of held context -- is amazing, and I believe will inspire some groups out there to put out open versions.
I love that the view count is included in the minimalist UI. I came across one with zero views, and there's something so intimate and exciting about being the first person to watch an ancient home video (even if it's shaky handycam footage of a horse, narrated in Russian).
As an aside, hats off to Google to being able to serve an 11 year old video with no noticeable delay from what must be the coldest of caches.
Phoenix creator here – excited to finally have shipped this! Happy to answer any elixir/phoenix/liveview questions.
In case folks missed it, buried in the blog post is a new installer that lets folks try out elixir/phoenix in seconds. It installs elixir and generates a new phoenix project from a single command:
osx/linux:
$ curl https://new.phoenixframework.org/yourappname | sh
> curl.exe -fsSO https://new.phoenixframework.org/app.bat; .\app.bat
edit: You can see it in action here: https://x.com/chris_mccord/status/1864067247255306332
Of course we also have non |sh installation guides if that's what you're after: https://hexdocs.pm/phoenix/installation.html
Now that this is out, I'm looking forward to put together a few new demos. What would folks like to see? Happy hacking!
Really resonated with this, reminded me of the journey I went on over the course of my dev career. By the end, my advice for every manager was roughly:
* Don't add process just for the sake of it. Only add it if seriously needed.
* Require ownership all the way to prod and beyond, no matter the role. (Turns out people tend to really like that.)
* Stop making reactive decisions. If something bad happened on a total, extremely unlikely lark, don't act like it's going to happen again next week.
* Resist the urge to build walls between people/teams/departments. Instead, build a culture of collaboration (Hard and squishy and difficult to scale? Yup. Worth it? Absolutely.)
* Never forget your team is full of actual humans.
Disclosure: HPC admin who works with NIVIDA cards here.
Because, no. It's not as simple as that.
NVIDIA has a complete ecosystem now. They have cards. They have cards of cards (platforms), which they produce, validate and sell. They have NVLink crossbars and switches which connects these cards on their card of cards with very high speeds and low latency.
For inter-server communication they have libraries which coordinate cards, workloads and computations.
They bought Mellanox, but that can be used by anyone, so there's no lock-in for now.
As a tangent, NVIDIA has a whole set of standards for pumping tremendous amount of data in and out of these mesh of cards. Let it be GPU-Direct storage or specialized daemons which handle data transfers on and off cards.
If you think that you can connect n cards on PCIe bus and just send workloads to them and solve problems magically, you'll hurt yourself a lot, both performance and psychology wise.
You have to build a stack which can perform these things with maximum possible performance to be able to compute with NVIDIA. It's not just emulating CUDA, now. Esp., on the high end of the AI spectrum (GenAI, MultiCard, MultiSystem, etc.).
For other lower end, multi-tenant scenarios, they have card virtualization, MIG, etc. for card sharing. You have to complete on that, too, for cloud and smaller applications.
Why don't they just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch? The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat. But beancounters running the company would never get such an idea...
Off topic but if I may. The way people use percentages to express multiples is confusing. A doubling is a 100% increase. So a 200% increase is 3x, and so on. Then at some point we forget about the +1 and 1000% is "10 times the sum it previously paid for software licenses". Just a pet peeve I guess : )
> This is the support email address you are looking for: [email protected]
Thank you. I have a bunch of curl tech support questions, but didn't know who to contact.
> it seems to me as a layman that the more radical proposition would be that the blood-brain barrier would be 100% effective at keeping out all bacteria,
Bacteria is not usually found in most organs in the body though. Even more so in an organ as important as the brain that even our own immune system doesn't have access. The brain has its own cells to do that job. In order for bacteria to cross the blood brain barrier you need bacteria in the blood first, where they aren't normally found without getting under fire.
Having bacteria in the gut is more obvious and makes more sense.
I kind of like Root's approach, just make three separate rulebooks. One quickstart guide, one "normal" rulebook, and one "law" type rulebook laying out everything in almost procedural style with clear and consistent definitions etc. Many games could benefit from that.
Once you try the demos, the animated image at the top feels misleading. Each segment cuts at just the right point to make you think you’d be able continue exploring these vast worlds, but in practice you can only walk a couple os steps before hitting an invisible wall, which becomes more frustrating than not being able to move at all. It feels like being trapped in a box. My reaction went from impressed to disappointed fast.
I get these are early steps, but they oversold it.
I feel like this is a mistake. Pat's strategy is aggressive but what the company needs.
Intel's stock is jumping at this announcement, but I look at it as a bad signal for Intel 18a. If 18a was looking to be a smash hit then I don't think Pat gets retired. If 18a is a success then it is an even more short-sighted decision by the board.
What this likely means is two-fold:
1. Intel 18a is being delayed further and/or there are significant issues that will hamstring performance.
2. Pat is/was unwilling to split the foundry and design business / be part of a M&A but the board wants to do one or the other.
If 18a is not ready I think the best case scenario for Intel is a merger with AMD. The US Govt would probably co-sign on it for national security concerns overriding the fact that it creates an absolute monopoly on x86 processors. The moat of the two companies together would give the new combined company plenty of time to ramp up their fabs.
For those not familiar with recent UK scandals, there has been one ongoing for a few years where there was a bug in a system designed by Fujitsu that was calculating the finances for post offices. This resulted in the system quoting wrong balances for post offices, and when that didn't match the actual cash in hand, resulted in prosecutions of the postmasters responsible for those offices which in turn led to convictions, dismissals, fines, imprisonments, and suicides over a 15 year period as the computer system was presumed correct.
Anyway, it was later proven the computer system was incorrect but the government there dragged their heels on exoneration and compensation.
https://en.wikipedia.org/wiki/British_Post_Office_scandal
I was a fine dining chef for 17 years. If you were in San Francisco during the 90s, I might have cooked for you. A simple way to increase demand is to remove or change the least popular menu item every week or every month. This technique has been so successful for me I wouldn't waste much time doing anything else. Once I made a Napoleon pastry for a desert special while at La Folie on Polk st. which sadly closed recently. I had extra pastry cream and made Paris-Brest because we never threw out food. One waitress sold Napoleon on every table and the other waitress sold nothing but Paris-Brest. The dishes were fundamentally the same so I made both anytime I did one or the other as a special because waitstaff for reasons not explained in the article will sell nothing but one and others the other. I made cheese dishes to sell before the desert and fruit soups in the summer. This is the mid 90s and we were tracking data. The chef pulled me aside and showed me the sales from the previous month because I sold 1.2 deserts per customer.
Nonetheless, for the last six years I cooked I was a private chef on a mega yacht. People ask me if the guests told me what they want to eat. I say never because I never asked what people want to eat. I cooked what I want to eat and then make enough for the guests and crew. It is the best menu strategy. In fine dining, the customers make decisions all day long and in the case of being a private yacht chef, the guests are making million and billion dollar decisions. The last thing they want to do is have to decide what to eat for dinner. The family I cooked for rarely ate off the boat. And when they did it was because I said something like, "I hear there is a very good restaurant in St. Barts named ....," which was code for "I want the night off."
I believe the reason one waitress would sell every last Paris-Brest and the other would sell every last Napoleon was because they told the guests in the restaurant what they wanted to eat for desert. They made the decision for the guests.
I know it's not public domain per-say, but for me, the thing that's most exciting is that in 2025, the last remaining patents on the h.264 (AVC) video codec will expire [1].
Now if only HEVC wasn't such a hot patent / licensing mess.
[1] https://meta.wikimedia.org/wiki/Have_the_patents_for_H.264_M...
Last year I got stuck on Day 12 for a full week, and thinking about how to solve it consumed my every waking moment. I think this year, I'm going to be kind to myself and not participate so I can really enjoy the winter break from work.