dgacmu

Thank you for finding this - the original is a really interesting article.

(@dang - consider re-pointing to this?)

No, I just pre-fill a cup with hot water from the steam wand (it has a hot water mode) and then pull a double shot of espresso into it.

(That's technically an Australian long black but I like it better - it leaves the crema on top.)

Most modern ethernet chips, including those used on USB ethernet devices, have adaptive interrupt coalescing (or moderation) for network I/O, which renders this likely not as big a deal as it once was. There will still be limits on packets/sec/core but it's not because of interrupts.

I finally gave in and got a breville bambino plus and I have no regrets. It was $400 on sale, produces quite decent espresso, foams milk well, has been very reliable, and doesn't heat up the kitchen - I held off getting an espresso machine for years because the instant heat ones always seemed to suck. This one doesn't. I mostly drink Americanos and straight espressos, my wife drinks lattes, everybody is happy.

But I have to agree with others: for my diluted espresso desires, I used (and still have) an Aeropress for years and it's simply fantastic. Low cost, almost zero maintenance, good results. Very similar output to an americano though lower on crema.

As others have noted, urban gas stations are likely to be far less busy in an EV world due to the ease of distributed charging -- home, work, destination, etc.;

But I think you raise a good model for long-haul. I think of the pennsylvania turnpike gas stations as a worst-case situation: They serve a somewhat captive audience, many of whom are traveling so far they need a mid-trip fillup. So something like 80kWh/minute _does_ seem like what you'd have to do for those specific stations, and that's an average rate of 4.8MW, at least during prime time.

You can probably get away with half of that if you use local storage as use is much lower at night. But let's not - let's see what it takes to do 4.8MW.

The answer is: You don't need a substation. You DO need on-site transformers and switchgear from 12kV primary service. But to put it in perspective, 4MW is like a tiny datacenter or really big (new york size) office building. So it's not really too crazy to think about an EV per minute going from 0-80kWh in a dedicated area. Compared to huge underground gas tanks, I think the infrastructure part of it is pretty ok.

What's nice there, though, is that the total amount of _energy_ needed at a charging station is roughly fixed(), regardless of how fast you charge the cars. So if you're provisioned for the needed total energy inflow, you can to a reasonable degree compensate for having a more bursty high-rate charging load by having some amount of local energy storage as a buffer.

() - Assuming you provision for the highest-traffic-volume day. Ignoring potential induced demand of making it a little easier to drive, which I suspect is pretty bounded - people need pee and stretch breaks anyway.

If you get 400km per charge using 88% of the battery (98% -> 10%), that's 400,000km (258,000 miles) before you're down to 90%, at which point you have likely worn out an awful lot of other things with the car.

Admitting that I have the luxury of an urban, low-driving lifestyle: I'm 50. That battery would literally last the rest of my driving life and have room to spare.

Ditto. I've mostly stopped using anchovies for everyday Caesars. The fish sauce trick is just too easy and tastes great.

I think it's less about blaming us for boycotting Tesla and more about blaming us for letting our entrenched interests in both oil extraction and ICE cars prevent us from investing in EV development and switching to them faster.

You can get them used for that price, and new for $107 if you buy qty 10+. See signature solar as one example.

Installation costs and inverters not included, however.

No provider maintains 100% utilization of GPUs at full rate. Demand is bursty - even if this project is successful, you might expect, e.g., things to be busy during the stock market times when Claude is throwing API errors and then severely underutilized during the same times that Anthropic was offering two-for-one off peak use.

And then there's a hit for overprovisioning in general. If the network is not overprovisioned somewhat, customers won't be able to get requests handled when they want, and they'll flee. But the more overprovisioned it is, the worse it is for compute seller earnings.

I suspect an optimistic view of earnings from a platform like this would be something like 1/8 utilization on a model like Gemma 4. Their calculator estimates my m4 pro mini could earn about $24/month at 3 hours/day on that model. That seems plausible.

@eigengajesh - Your cost estimator lists Mac Mini M4 Pro with only 24 or 48GB options, but the M4 Pro mini can also be configured with 64GB. At least, I hope so, as I'm typing this on one. ;-)

Oh, also, you seem to have some bugs:

Gemma: WARN [vllm_mlx] RuntimeError: Failed to load the default metallib. This library is using language version 4.0 which is not supported on this OS. library not found library not found library not found

cohere: 2026-04-16T14:25:10.541562Z WARN [stt] File "/Users/dga/.darkbloom/bin/stt_server.py", line 332, in load_model 2026-04-16T14:25:10.541614Z WARN [stt] from mlx_audio.stt.models.cohere_asr import audio as audio_mod 2026-04-16T14:25:10.541643Z WARN [stt] ModuleNotFoundError: No module named 'mlx_audio.stt.models.cohere_asr'

Trying to download the flux image models fails with:

curl: (56) The requested URL returned error: 404

darkbloom earnings does not work

your documentation is inconstent between saying 100% of revenue to providers vs 95%

I think .. this needs a little more care and feeding before you open it up widely. :) And maybe lay off the LLM generated text before it gets you in trouble for promising things you're not delivering.

Commonly used as a projectile

It is, but the other one is a link to their twitter post, whereas this is the longer self-hosted statement. This is a better, more informative source.

*naivete

that word and I have never gotten along well. insert obligatory joke at my expense.

That's why I said if asked in good faith. :-)

But seriously GP could have had a mental model that landfilled orange peels might sit there for a long time -- which depending on conditions and food could be true on human scales (like 10-40 years) but not on the scale of 100 years. Especially if the conditions were dry -- a dry orange peel is pretty robust. That's not likely to be the case in Costa Rica, but I'll forgive some naivety here absent demonstrated malice.

I don't. The batteries will last longer than 10 years. The 10 year typical advertised lifetime of lifepo4 is to 80% capacity, and I'll just keep on using them.

The actual payoff calculation is a lot messier than that because you have to factor in the NPV of buying batteries vs. just throwing the money in the market, AND you have to be able to forecast that growth vs. growth in power prices. So the honest truth is I have no idea if it's going to be a net good investment vs other options.

Fortunately, I don't have to care, because I bought the batteries for UPS runtime, which I value independent of the time-shifting. The time-shifting is just a way to squeeze money out of an investment I already made. Had I been going for payoff, there are cheaper battery/inverter options out there with a sub-5y payoff.

You're getting downvoted but it's a reasonable question if posed in good faith. The tl;dr is that there are really a few options for what could happen to those orange peels:

(1) Landfill burial

   (1a) Without methane capture and use: Produces methane, relatively high short term warming potential.
   (1b) With methane capture and use: Ends up as CO2 after burning the methane.

(2) Composting (this approach)

   (2a) Mostly aerobic: Produces CO2
   (2b) Mostly anaerobic: Produces methane

A deep pile that is never turned will decompose anaerobically, resulting in fairly undesirable methane. A shallower pile or one that is mixed well will result in mostly aerobic decomposition. The aerobic decomposition will produce CO2 but not huge amounts of it. Each hectare of land could absorb something like ~8 tons of CO2 per year; with 7 hectares, the CO2 emitted by composting 12t of oranges is going to be dwarfed by the new vegetation. After a few years when you're growing big trees, the rate of CO2 absorption might rise as high as 20-30t/year/hectare in costa rica's environment. And this is probably an underestimate, as the soil amendment of the orange peels seems to have stimulated faster regrowth than would have happened otherwise.

And perhaps more to the point: There isn't really a purely "no co2" way of disposing of organic matter other than perhaps burying it at the bottom of a deep mineshaft (but the co2 or methane will be produced anyway). Landfilling it is strictly worse - you still get the decomposition products, _or worse_ because you'll mostly get methane, but without producing useful soil byproducts.

Overall this project is a huge win on a carbon perspective and a waste reduction perspective.

I don't sell mine, but I time-shift with a small pile of batteries (about 10kWh) and it's pretty reasonable. I save about usd $30/month. It's basically a big ups that will pay for itself in ten years, and I get backup power.

The pile of shell and sed is cleaning up the ai output and then running it in the shell.

The instruction to the AI was to create _a_ shell command. So it's a random shell command generator (maybe).

I find it kind of helpful and interesting to see a subset of these called out with a bit of data. Helps keep my LLM detector trained (the one in my brain, that is) and I think it helps a little about expressing the community consensus against this crap. In this case, I'm glad the GP posted something, as it's definitely not mistaken.

As a specific example: The generated diagram showing the expression tree under "build in python" is simply wrong. It doesn't correspond to the expression x * 2 + 1, which should have only 1 child node on the right. The "GIL Released - Released" is just confusing. The dataflow omits the fact that the results end up back in python - there should be a return arrow. etc., etc.

If you use diagrams like this, at least ensure they are accurately conveying the right understanding.

And in general, listen to the person I'm responding to -- be really deliberate with your graphics or omit. Most AI-generated diagrams are crap.

It's written in rust, but why do you believe it was co-authored with Claude? The README in github specifically says:

> This project was built using Gemini CLI

https://github.com/sashiko-dev/sashiko

An enthusiastic two thumbs up to this approach. It's exactly what I run at home that has been working solidly. I run on an N100, which is just a hair smaller than an i5-8500, with 32GB DRAM and a 1TB SSD (total overkill). I keep it under proxmox; the box also runs my unifi SDN controller, pihole, and a linux VM for various little services. Two USB dongles for z-wave / zigbee / matter (because I'm a glutton for punishment). Backed up to a NAS. It's fast, easy, and has been very reliable.

Mass-market paperbacks are definitely dying, but trade paperbacks continue to sell (at rates lower than mass-market, obviously):

https://www.publishersweekly.com/pw/by-topic/industry-news/p...

(trade paperbacks are the larger paperback editions printed on better paper than the mass market paperbacks, but still soft-cover.)

John Scalzi posted about this a few months ago:

"All my recent books went from hardcover to trade paperback and almost all of my backlist in mass market has now migrated to trade. The role of mass market paperbacks is now handled almost entirely by ebooks."

https://bsky.app/profile/scalzi.com/post/3m7xzfxxcg222

Gemini 3.1 pro under a Google AI pro subscription has just recently started imposing really small weekly limits. I went from it feeling unlimited to hitting a 4 day quota in 2 hours of use. Very odd. Wonder if too many people jumped on with the 3.1 pro release.

Something I try very hard to impress on my PhD students is that the process of writing is part of the process of thinking. We often have cool things in our head that don't sound right when we write them down, and that's usually because the thing in our head was more amorphous than we realized. The time you put in getting the written expression of it to work is actually helping you crystallize what you're thinking in the first place.

Would anyone notice if you spell-checked or got narrow feedback about grammar? No. I'm not dang, but perhaps a very reasonable interpretation of the rules is: If the AI is generating the words, don't. If it tells you something about your words and you choose to revise them without just copying words the AI output, it's still your words.

(As an experiment, I took that paragraph and threw it into gemini to ask for spell and grammar checking. It yelled at me completely incorrectly about saying "I'm not dang". Of its 4 suggestions, only 1 was correct, and the other 3 would have either broken what I was trying to say or reduced the presence of my usual HN comment voice. So while I said the above, perhaps I'm wrong and even listening to the damn box about grammar is a bad idea.)

That said, I often post from my phone and have somewhat frequent little glitches either from voice recognition or large clumsy thumbs, and nobody has ever seemed to care except me when I notice them a few minutes after the edit button goes away.

Shipping from China to a US coastal warehouse is probably modestly under $200, including packaging, assuming it's shipped in a 40" container. Possibly less if there's other cargo that can be used to fill the remaining space.

I suspect the domestic costs are really dependent on volume (like, can you ship a container of 45 TVs to a warehouse near NYC or do you have to ship each unit individually) and I don't feel confident estimating that side of it.

I don't disagree.

I assume that most of these purely llm generated unwanted contributions will just end up in dead end forks, because my impression is that a lot of them are just being generated as GitHub activity fodder. But the stuff that really solves a problem for a person - eh, good. Problem solved is problem solved. (Unless it creates new problems)