mattlutze

>Also, nobody really objects to the research that leads directly to stuff private industry can use. That's not what people want to cut.

That's what they're cutting.

The expectation that one model be top marks for all things is, imo, asking too much.

Too many people think these are funds to just run the universities. By and large, what is being withheld are funds for research.

Federally-funded academic science often looks like:

  1. The university + government fund/run a project
  2. Project creates new knowledge (cool!)
  3. The government gets a pretty awesome license to use that knowledge
  4. The government more often than not gives that knowledge away (or offers great accessible licensing), so that
  5. Private industry can adapt, apply and commercialize the knowledge, driving new GDP growth and opportunities for improving life, etc.

Withholding these funds ends the research projects, because Universities are not startup incubators. So the research stops, and one of the highest returning pipelines of new GDP growth for the US dries up—unless today, the professors and universities kiss the president's ring and promise to wipe out 50-100 years of human rights improvements.

Honestly I'm here for it, because it's an option for a market of groups that don't otherwise have the opportunity to deploy this kind of capability. Teams feelings aside :)

I worked for a client once that refused to let us build and manage databases for things that needed it. The one option in the end that we could get approved was using Microsoft SharePoint lists and CRUD'ing to them through the Javascript API.

A lot of problems have lame constraints, but having an option at all to solve them is pretty nice.

From a very inhuman perspective, and one I don't find appropriate to generally use: A human physical worker is a high capital and operational expense. A robot may not have such high costs in the end.

Before a human physical worker can start being productive, they need to be educated for 10-16+ years, while being fed, clothed, sheltered and entertained. Then they require ongoing income to fund their personal food, clothing and shelter, as well as many varieties of entertainment and community to maintain long-term psychological well-being.

A robot strips so much of this down to energy in, energy out. The durability and adaptability of a robot can be optimized to the kinds of work it will do, and unit economics will design a way to make accessible the capital cost of preparing a robot for service.

Emotional opinions on AI aside, we will I think see many additional high-tech support options in the coming decade for physical trades and design trades alike.

Customer service, entry sales, jr data/business specialist

- Ada's LLM chatbot does a good enough job to meet service expectations.

- AgentVoice lets you build voice/sms/email agents and run cold sales and follow ups (probably others better it was just the first one I found)

- Dot (getdot.ai) gives you an agent in Slack that can query and analyze internal databases, answering many entry level kinds of data questions.

Does that mean these jobs at the entry level go away? Honestly probably not. A few fewer will get hired in any company, but more companies will be able to create hybrid junior roles that look like an office manager or general operations specialist with superpowers, and entry level folks are going to step quickly up a level of abstraction.

For $32B Google are buying Wiz's brand, existing customers and their pipeline of customers, along with the technology.

I was wondering why like every Wiz business development person was cold-engaging me on LinkedIn and email last year.

Russia broke treaty, law, moral foundation and democratic order for a second time in a decade to try and take all of the Ukraine.

You appear to forget history, but plenty of citizens were able to leave through the back door while able-bodied citizens were required to do the only reasonable thing a democracy can do and defend its sovereignty.

First with what they had, then with conservatively provided weapons as other countries tried to figure out if Putin would actually use nukes, and finally with more effective munitions when it was apparent he would not.

No neighbor is locked in, and the US embarrasses itself as it simps for Russia in the UN and parades the president of an invaded democracy around in a jester hat.

I would love to provide recommendations and help with menu ideas. If you have a look in my profile you can get to my address and email me from there <3

My partner and I have been fully plant based since 2018 and variously play rugby and train triathlons.

Significant input costs per animal calorie means there's an order of magnitude or more insects being killed to raise the animal that you ultimately end up eating, to get the same calories you'd have gotten just eating the plant sources.

Harvesting crops is materially different from slaughtering animals, and calorie for calorie, plant-based nutrition involves less termination of life than getting calories from animals (if you're grouping insects and non-animal life into the "forms of life" being killed).

If people don't like killing in general, or killing animals more specifically, they can live a wonderfully health(y|ier) life by going plant-based, be responsible for less killing, and today do it without having to give up the textures and experiences they've be conditioned on.

It's difficult in 2025 to conclude that a person who doesn't choose to eat this way is particularly opposed to killing, in the way that you propose.

There's a few that have tried to implement this, and I want to live in none of them.

The US will fare no better if it walks down this path, and honestly will likely fare worse for it's cultural obsession with individualism over community.

Tax breaks, i.e. a company extracting wealth from a community without paying into the systems that keep all the parts of that community running, forcing the community to ultimate subsidize that business's weath extraction from them.

Ellison should be nowhere near this:

https://arstechnica.com/information-technology/2024/09/omnip...

The man has the moral system of a private prison and the money to build one.

Go low-tech and start printing a small local newspaper.

Pay for it with ads from local businesses, and give it away for free at all those stores. Get your regional Chamber of Commerce to help set you up with connections and sales channels.

> Larger companies will often write you off as being "not experienced" enough which is often code for "you don't look old enough for this position".

It's not about age, it's about what you actually had to do. Company size, company age and political complexity are strongly correlated.

A successful large company engineering director spends their day mostly doing politics across the org, while a successful small company engineering director spends a lot of their day writing code, being an architect, filling in for product management.

> Unlike engineering and the startup world, big companies want you to look the part not just know the part.

Big companies want you to know how to play the long game.

For-profit companies don't have to publish papers on the SOTA they product. In previous generations and other industries, it was common to keep some things locked away as company secrets.

But Google, OpenAI and Meta have chosen to let their teams mostly publish their innovations, because they've decided either to be terribly altruistic or that there's a financial benefit in their researchers getting timely credit for their science.

But that means then that anyone with access can read and adapt. They give up the moat for notariety.

And it's a fine comparison to look at how others have leapfrogged. Anthropic is similarly young—just 3 and a bit years old—but no one is accusing them of riding other companies' coat tails in the success of their current frontier models.

A final note that may not need saying is: it's also very difficult to make big tech small while maintaining capabilities. The engineering work they've done is impressive and a credit to the inginuity of their staff.

One of my soapboxes is that the world is full of really smart people trying to solve solved problems for the sake of having done something that feels like something. I advocate for people taking that creative energy and pointing it at partially/poorly/un- solved problems.

If that's developing to offer (OP in particualrly said their friend would soon be offering a new service) an RSS Reader because there's something to actually innovate and more fully solve a problem, then it's great. Otherwise, it's just fodder for AdWords and Facebook when they eventually try to build acquisition for a business around it.

> At its core, the PC movement was about a kind of tech liberty—–which I’ll define as the freedom to explore new ideas, control your own creative works, and make mistakes without punishment.

The PC movement of the 90s, where it feels like this author is reminiscing, was about arbitraging the delta between what the tech could do and the literacy and expertise in government.

> But over the past decade in particular, the Internet and digital rights management (DRM) have been steadily pulling that control away from us and putting it into the hands of huge corporations.

This period of computing was notable for how a bunch of nerds figured out how to use new networking technology to stretch/abuse/violate/break copyright and fair-use laws around media.

So many ways to get ripped content then. It was fun for a teenager and felt like a victimless loophole. It both opened a bloom of interesting new creative works, but also decimating existing markets and systems, so that eventually new monopolists Netflix and Spotify could take over.

But the conditions and tools available then never went away, and private personal computing is more available today than ever before. In a few hours someone can read a few tutorials and buy/build and run a whole redudant content serving service for all of their personal needs, while writing or conglomerating a whole system of tools and capabilities to automate or agument nearly anything they could think of.

I changed phones, and tried to log back into my Fediverse/Mastadon accounts. The app asks me which servers I'm on—I can't find the accounts in my password manager, can't figure out which servers they were, and the ones I thought I was on maybe don't exist anymore? Or were accessible in one but not another app.

So I managed to log into one of the 3 accounts I'm sure that I have still. And I'm a software nerd who makes "educated" decisions all the time around this stuff.

Protocol People really care about that, and you know what? It becomes their network effect. But it is a self-selecting network. The nature or design of what effects and attracts the network is the same mechanism for limiting its size.

TikTok, Instagram, Snapchat all focus on things that other people really care about—namely video creation, photo curation and ephemeral small network cohesion. and those focuses attract other userbases.

Probably, there's a lot more people who want to create and watch short videos than there are people who want to nerd out over what their 1/10,000 servers' community rules and protocol settings are.

Does the world need another RSS reader/mousetrap? We already have so many.

https://duckduckgo.com/?t=ffab&q=rss+readers

What is your friend's idea to revolutionize with their new reader?

https://artificialintelligenceact.eu/high-level-summary/

There's many places where the model might be used which could count as high-risk scenarios and require lots of controls. Also, we have:

  GPAI models present systemic risks when the cumulative amount of compute used for its training is greater than 10^25 floating point operations (FLOPs). Providers must notify the Commission if their model meets this criterion within 2 weeks. The provider may present arguments that, despite meeting the criteria, their model does not present systemic risks. The Commission may decide on its own, or via a qualified alert from the scientific panel of independent experts, that a model has high impact capabilities, rendering it systemic.

  In addition to the four obligations above, providers of GPAI models with systemic risk must also:

  - Perform model evaluations, including conducting and documenting adversarial testing to identify and mitigate systemic risk.
  - Assess and mitigate possible systemic risks, including their sources.
  - Track, document and report serious incidents and possible corrective measures to the AI Office and relevant national competent authorities without undue delay.
  - Ensure an adequate level of cybersecurity protection."

They may not want to meet these requirements.

If teams are hiring Junior developers to give them "grunt work," they're doing something wrong to begin with.

Unique datasets are the moat for AI businesses. The data from these glasses is quite novel in it's perspective, context and contiguous-ness among other attributes.

Versus mobile phones, Meta are making a better go at, and seem to be the leader on, the bet that fundamentally ML/AI-driven glasses are going to be the next default modality for UX on the Internet.

I think you're describing one of the core motivations for these decades of research.

In most places still, changes to these maps need to go through lots of people process. This implies they're slow to update, and therefore often not entirely accurate or optimally understandable. For example, consider service works or temporary outtages.

The outcome of this research is not a toy-prize for armchair readers, but a generalized algorithmic approach to building necessary and important accessibility tools.

From the introduction,

  Since the days of Harry Beck, transit maps have mostly been created manually by professional map designers (Garland Citation1994; Wu et al. Citation2020). The primary focus was on static maps, either distributed in print or electronically. These maps are typically schematic, and the classic octilinear design (network segment orientations are multiples of 45∘) is still prevalent. In the late 1990s, the graph drawing community started to investigate the problem of drawing such maps automatically. The following questions were investigated: (1) How can graphs be drawn in an octilinear fashion? (2) Which hard criteria should a transit map fulfil? (3) Which soft criteria should be optimized? Several methods have since been proposed (see below). A set of soft and hard criteria, first described by Nöllenburg (Citation2005), has since been generally accepted. The important sub-problem of finding an optimal line ordering of lines travelling through network segments has also been identified very early by Benkert et al. (Citation2006).

It not trivial to automatically generate an optimally understandable octalinear transit map, and this group have combined bits of 30 years of research to do it in one go for every* bit of public transit on the planet.

* every bit that's in OSM, I suppose

If they don't call home, then the foreign interests wouldn't be getting the data, I think? If the concern is access to inferred logistics and crane telemetrics, then the solution would still seem to be removing access to the data or retrofitting new controls systems to the cranes, and not replacing the entire crane.

Why do we not just remove the RFID radios and cameras from the cranes, or just not allow them to call home?

This is a problem that doesn't require billions in investment to solve, but rather a wire cutter.

If someone uses an LLM to produce the code, I'd guess they'll use it to evaluate the code as well.