Another reason to favor using AI to build automation instead of relying on it in prod: the risk of war and global instability.
If LLMs are genuinely helpful or even decisive in a military engagement, you can expect any host country to commandeer whatever data centers they need, leaving commercial entities to bid up the prices on the leftover capacity.
Another risk is that data centers are a great target for cyber warfare.
It’s ideal if your business can leverage LLMs when they’re online but continue to operate profitably when they’re offline.
Even regular warfare, if the Middle East AWS regions are an indication. The giant and arguably excessive data centers being built are not hardened physically.
Putting the generic term into your corporation's name can be effective means of claiming things that don't belong to you.
Jon Postel reserved 44.0.0.0/8 for a generic purpose: "amateur radio digital communications." Decades later, there was a successful heist when some enterprising individuals who had incorporated "Amateur Radio Digital Communications" misrepresented to ARIN that the assignment had actually been theirs. Immediately after ARIN gave them transfer rights, they pocketed 8 figures reselling the space to Amazon.
Github obviously isn't making explicit claims like this but they benefit whenever people with purchasing power implicitly understand that github is the only option.
edited: Amateur Radio Digital Communications is not an LLC
This lengthy email thread[0] indicates that Jon Postel made the assignment in 1992, that the entity "Amateur Radio Digital Communications" wasn't formed until years later, meaning Jon's assignment had to have been for a purpose and not to an entity of the same name.
The head of ARIN defends[1] the transfer throughout the thread.
From your earlier comment it sounded like there was a "heist" simply based on having a similar name. Looking into it though, it seems like the ARDC non-profit did a pretty reasonable job of proving they were the same folks who'd been managing the IP block for decades. Also, has there been any sort of allegation that they've misused the funds? From what I understand they've pretty consistently used the funds to support amateur radio.
Just out of curiosity, are you familiar with infrastructures.org? It's a manifesto on how to bootstrap infrastructure from a team that would go on to invent IaC tooling. I read the site thoroughly a few years back and wondered how much differently you'd order the bootstrapping process for modern infrastructure.
I have been considering Nextcloud for a homelab install, but dreading the deployment/maintenance story there if I ever start relying on it as a source of truth. Would definitely consider a smaller option like this.
My biggest worry is your supply chain. If you're mostly using AI, would love to see you build your own library of functions and drop as many direct and transient dependencies as possible.
hah, so I've been fighting AI all the time as it tries to write all this on it's own vs re-using libraries. To me that's a much safer option, because the libraries all pre-date AI so I assume the authors know what they're doing.
Not to bag on Nextcloud but I do think that this is _much_ easier to understand and maintain, it's standard Go/React/Sqlite, nothing esoteric so if you did need to maintain it I'm sure anyone could pick it up. Of course I don't really know Php so that def flavors my opinions here.
The claims you make about Go/React are similar to the claims made by PHP and Python and JS users. Basically: the thing I'm familiar with is easy - then imply that it would be the same level of easy for others.
This does look easier than Nextcloud however I'm not sure thats the tech-stack. Generally the new project is easier/simpler because it doesn't have so many years of scope-creep, work-arounds, half-imolemented-then-abandond features.
yep very true, everyone has their preferred stack, and greenfield is _always_ easier than maintaining years worth of past code.
I did look really closely at Nextcloud when it came time to leave Google apps. I really wanted it to work but also wanted:
* Realtime updates - Pockebase gives us this for free.
* A native app - React Native <- this one worked out but does cause pain with the web version, some things don't work _quite_ as well as a standard React ap would.
* Lightweight docs that work on mobile. Collabora is very full-featured but is also pretty slow. It remains to be seen if I've successfully here because we're only starting to use our docs "for real"
And also: I'm absolutely sure this has many bugs that I've yet to uncover. All I can say is that I am planning on using it for real-world business stuff so I'm sure we'll be uncovering and fixing them throughout the next year.
Not all PE problems are existential; they will be outcompeted.
What keeps a newly graduated Veterinarian from opening her own clinic and undercutting the PE competition? With no massive loans on her books, she can profitably offer lower prices than PE can. She may even drive the local PE clinic out of business.
First, opening a clinic requires some serious money. Then, if the new clinic gets traction, PE can make a very good offer for a buyout and the owner would have to be stupid or very stubborn to refuse. Most big companies these days just buy up competition. Good for the owners but bad for the customers.
Another thing which I've seen personally is that in general getting from 0 to 1 required HEROIC effort to the point that it requires a personality type.
How many veterinarians got into the game to become relentless, driven, scrappy and indomitable business owners vs because they love furry things and helping?
Except every newly-graduated veterinarian does have a massive loan on their books, in the form of student loans. And even if she didn't, where does the startup capital for her clinic come from? Whether in human or animal medicine, starting your own practice--especially as a new grad--is usually the course of action with the highest-risk-to-lowest-pay ratio.
> With no massive loans on her books, she can profitably offer lower prices than PE can
Depends entirely on fixed vs variable costs. Rollups (which are very common now) work mainly because most "mom and pop" businesses can easily be "unlocked" by pooling the treasury, HR, accounting, commercial banking, supplier negotiations etc.
People had no problem starting clinics in the past and they probably could today if they really wanted to, but there is little incentive to for the past few decades while the stock market has been booming. Why spend 80 hours a week struggling to run your own clinic and paying off loans when you can work 40 hours a week working for somebody else's clinic and invest your savings in the stock market?
Assuming you had $$$ for some supplies but couldn't afford to lease a commercial building, you could provide small mammal services from your vehicle, driving to people's homes to give vaccinations and well care.
Being mobile would also allow you to serve a larger market than a fixed clinic; you could serve a couple small towns on Monday, a couple others on Tuesday, and server a larger metro on the weekends.
Once you're consistently profiting $$$$/day you'll be able to start saving for the equipment you'll need for a commercial lease somewhere because you have both the cash, cash flow, a loyal customer base, and critically, a good sense of where a good location would be to serve them from.
Sure, that sounds plausible. I'm not saying you need an enormous amount of money, but for this scenario you need supplies, car payments, gas, probably some kind of licensing fee, insurance, some kind of advertising, and a few months of rent and living expenses until you start making a profit. Maybe like $10,000, plus more as a safety net in case it doesn't work out and you need to find a job?
Even if they are lucky enough to have no debt, I don't think the average graduate has $10,000+ in the bank to spend. I have never started a business so I honestly have no idea how hard it is to get a small business loan for something like this, maybe it's easy, but even so it's certainly risky.
Yet there is no evidence of this happening in any industry or area where PE has become the dominant player. Why not? What you’re saying is nice economic theory but it’s clearly not happening.
How would you phrase this though? Plenty of PE firms have the funds to buy your local veterinary clinic or auto body shop with cash; the leverage comes later, when they direct the business that they own to get a loan. How can you make it illegal for the business to get a loan?
The daughter company would presumable be allowed to purchase goods and services. What prevents those goods and services from being supplied (at a hefty markup) by another company under PE control?
The US Federal Government spends twice as much as it collects in taxes, borrowing the difference. The mounting debt requires that we set interest rates as low as possible to keep interest payments feasible.
For depositors, this means you can't make money in the bank. And the stock markets gains look good on paper but inflation erases much of the real value. So people with giant pools of capital have learned to make their own fortunes by buying companies directly. This is "private equity."
Their playbook once they do so is limited to a few extractive techniques. They might buy a few leading competitors in an industry and merge them, double/triple the rates, and shutdown the associated 3rd party services "marketplace" and force people to buy only their services. Or start charging for API access that previously offered to all customers for free.
They might buy a service provider who charges reasonable rates, double/triple the rates, then sell them off again 14 months later.
They might buy a solvent company, saddle it with debt, and sell it off.
These private equity gains drive everyday costs for consumers like me. In a recent 24 months period, every monthly bill I pay went up $$$ as PE firms took over my service providers.
We could slow PE (and inflation in general) by raising interest rates, incentivizing deposits and increasing the cost of capital. But this would require national fiscal responsibility, and nobody wants that. Additionally, we could choose to bootstrap companies with sustainable multigeneration succession planning instead of sudden financialized cash outs. But after tirelessly building a company for a decade most founders would rather cash out so someone else can begin to abuse their customers. "I deserve this."
their point, if I distill it correctly, is that Meehan comes with an agenda of enshittification. Given the events we're talking here... it's not that far-fetched.
ok and it's worth discussing on its merits. but "wikipedia is private equity" is a dumb thing to say and it matters to actually make concrete criticisms instead of throwing around buzzwords. the agenda of private equity is driven by profit incentives that make absolutely zero sense here.
> the agenda of private equity is driven by profit incentives that make absolutely zero sense here.
well, that also makes a bit of sense. like, why focus on cashing in from AI crap or on continuously expanding the cash bunker like a PE firm would, instead of actually investing money into coordinating with the communities when it comes to stuff like a new theme?
> Once you know what all the people in a top publishing company do, the difference between an amateur publication and a professional one becomes immediately apparent.
Any advise for developing this sense?
I will never work in a top publishing company but I have been able to approximate good design by first studying the fundamentals, then reproducing the layouts I see in popular media. I can make text into a beautiful book, and I see poor design choices in the corporate communication billion dollar companies.
But it feels like there’s a lot more I don’t know, and you never know what you don’t know, and it makes me wish I could absorb more from working under an expert.
There's no substitute for apprenticeship (by whatever name). Unfortunately, skills of this kind may be close to extinction. For someone like you just interested in getting better at layout design, I'd recommend something like 'The Elements of Typographic Style', by Bringhurst; this concentrates mostly on books, but much applies to other layouts. Of more general interest -- i.e., beyond layout design -- might be 'An Encyclopedia of the Book', by Glaister. There's a wealth of valuable design and print resources from the '60s - '90s if you can find them -- some libraries still have high-quality examples, but most have replaced them with much less-valuable contemporary resources. Look for book and magazine sales by university departments, businesses, etc.
Thank you! I have been absorbing Bringhurst methodically the past year.
I had not heard of Glaister, will be on the lookout.
Good point about library and corporate sales. My main supply of materials from the 60s has been from estate sales -- not for instructional materials, but for well composed period pieces. Older letterfaces and color palettes are so evocative; seeing the label of a 70 year old oil can with so much more personality than the products of today makes me want to bottle this style for my own future use. And it feels good to hold something back from the landfill.
The trouble with our age is that, despite the abundance of intermediate-level information, expert teachers in specific, and shrinking, professions are as hard as ever to access, if not more so.
> some of the most meditative and creative of my life
This sounds like a worthy pursuit. We control the most powerful machines to ever have existed, yet it's all too easy to use them for anesthetic distraction. Offline relationship and meditation help develop our capacity to use these machines for something better.
The cost of the tooling is probably blocking a few sales, but the biggest blocker for FPGA sales is price of the parts themselves. FPGAs make the most sense where absolute quantities are low and the customer is not cost-sensitive.
As soon as you get volume, ASIC wins out. And cost sensitive applications can almost always make do with a CPU that has the correct IO configuration. For FPGAs to win at scale they'd need to be significantly cheaper.
If LLMs are genuinely helpful or even decisive in a military engagement, you can expect any host country to commandeer whatever data centers they need, leaving commercial entities to bid up the prices on the leftover capacity.
Another risk is that data centers are a great target for cyber warfare.
It’s ideal if your business can leverage LLMs when they’re online but continue to operate profitably when they’re offline.
reply