tal

Forgejo, codebergs self hosted option requires too much manpower/manual work and/or custom ansible work compared to others

Ehh. I mean, I haven't looked at it, but honestly, how hard would it be for someone to improve that?

searches

https://github.com/s-hamann/ansible-forgejo

I mean, if you're okay using that, the example per-site configuration there is like...36 lines of YAML.

EDIT: Oh, the example doesn't set up TLS, and I guess probably most people would want that.

I don't particularly have a problem with Codeberg or with GitHub, but if I were doing a software project and it were GPL-licensed and I didn't have an existing presence anywhere, I'd probably look at the FSF's Non-GNU Savannah. I have a pretty high level of trust that anything that the FSF does is going to be okay (and, frankly, it's one of the few organizations where there is necessarily a fair bit of trust in that the FSF can create updated versions of the GPL that GPL-licensed software can be used under anyway), and they've been running that forge for...checks...24 years now. I would bet that 24 years from now, it'll probably be more-or-less the same.

EDIT: I would add that it'd maybe be kind of interesting to have some kind of distributed forge. I mean, git is intrinsically distributed, but as I talked about in a recent discussion, issue-tracking isn't (though there is a distributed git-bug that I discovered in that discussion), and while you can do code review via email (and email is distributed), a lot of people use non-distributed Web frontends on forges for it.

Like, I think that it might be nice to just make forges distributed, make people not rely on centralized infrastructure in general.

If you are a chemist tasting stuff to identify properties is a bad idea.

I was appalled, reading through these:

1879:

https://web.archive.org/web/20170314015912/https://books.google.com/books?id=f4I9AQAAIAAJ&pg=PA36#v=onepage&q=&f=false

"How did I discover saccharine?" he said. "Well, it was partly by accident and partly by study. I had worked a long time upon the compound radicals and a substitution products of coal tar, and had made a number of scientific discoveries that are, as far as I know, of no commercial value. One evening I was so interested in my laboratory that I forgot about supper until quite late, and then rushed off for a meal without stopping to wash my hands. I sat down, broke a piece of bread, and put it to my lips. I tasted unspeakably sweet. I did not ask why it was so, probably because I thought it was some cake or sweetmeat. I rinsed my mouth with water, and dried my mustached with my napkin, when, to my surprise, the napkin tasted sweeter than the bread. Then I was puzzled. I again raised my goblet, and, as fortune would have it, applied my mouth where my fingers had touched it before. The water seemed sirup. It flashed upon me that I was the cause of the singular universal sweetness, and I accordingly tasted the end of my thumb, and found that it surpassed any confectionary I had ever eaten. I saw the whole thing at a glance. I had discovered or made some coal tar substance which out-sugared sugar. I dropped my dinner and ran back to the laboratory. There, in my excitement, I tasted the contents of every beaker and evaporating dish on the table. Luckily for me, non contained any corrosive or poisonous liquid.
"One of them contained an impure solution of saccharine..."

1965:

https://en.wikipedia.org/wiki/Aspartame#History

Aspartame was discovered by accident[60] in December 1965[61] by James M. Schlatter, a chemist working for G.D. Searle & Company in Skokie, Illinois. Schlatter had synthesized aspartame as an intermediate step in generating a tetrapeptide of the hormone gastrin, for use in assessing an anti-ulcer drug candidate.[60] He discovered its sweet taste when he licked his finger, which had become contaminated with aspartame, to lift up a piece of paper.[12][62]

1967:

https://en.wikipedia.org/wiki/Acesulfame_potassium#Discovery

Acesulfame potassium was developed after the accidental discovery of a similar compound (5,6-dimethyl-1,2,3-oxathiazin-4(3H)-one 2,2-dioxide) in 1967 by Karl Clauss and Harald Jensen at Hoechst AG.[16][17] After some of the chemical powder accidentally spilled onto his sweater, Clauss licked his finger to wipe the spots away, which led him to notice its intensely sweet taste.[18][19]

1975:

https://en.wikipedia.org/wiki/Sucralose

Sucralose was discovered in 1975 by scientists from Tate & Lyle, working with researchers Leslie Hough and Shashikant Phadnis at Queen Elizabeth College (now part of King's College London).[21] While researching novel uses of sucrose and its synthetic derivatives, mixing sulfuryl chloride with sugar, Phadnis was told to "test" a chlorinated sugar compound. According to an anecdotal account, Phadnis thought Hough asked him to "taste" it, so he did and found the compound to be exceptionally sweet.[22][23]

If Google and Apple start doing this wirh Bluetooth and WiFi on Location Services, they're not that far off the The Dark Knight sonar scene:

https://www.youtube.com/watch?v=IRELLH86Edo

double checks

It's not in there. Could be that the subspecies classification was recent or undone or controversial or something?

Bison bison bison

Oh sweet buttery Jesus yes.

https://en.wikipedia.org/wiki/Buffalo_buffalo_Buffalo_buffalo_buffalo_buffalo_Buffalo_buffalo

Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo.

Buffalo Bison bison bison Buffalo Bison bison bison buffalo buffalo Buffalo Bison bison bison.

        $ wget https://www.itis.gov/downloads/itisMySQLTables.tar.gz
    $ unp -U itisMySQLTables.tar.gz
    $ cd itisMySQL072826
    $ cut -d'|' -f2 longnames|grep -iE '^([^ ]+) \1 \1$'
    Porites porites porites
    Volva volva volva
    Modiolus modiolus modiolus
    Quadrula quadrula quadrula
    Arcinella arcinella arcinella
    Echiurus echiurus echiurus
    Sprattus sprattus sprattus
    Lepadogaster lepadogaster lepadogaster
    Pungitius pungitius pungitius
    Caretta caretta caretta
    Puffinus puffinus puffinus
    Sula sula sula
    Anhinga anhinga anhinga
    Nycticorax nycticorax nycticorax
    Lagopus lagopus lagopus
    Perdix perdix perdix
    Limosa limosa limosa
    Alle alle alle
    Apus apus apus
    Riparia riparia riparia
    Troglodytes troglodytes troglodytes
    Cardinalis cardinalis cardinalis
    Coccothraustes coccothraustes coccothraustes
    Pyrrhula pyrrhula pyrrhula
    Carduelis carduelis carduelis
    Pica pica pica
    Oenanthe oenanthe oenanthe
    Ciconia ciconia ciconia
    Bison bison bison
    Clelia clelia clelia
    Phocoena phocoena phocoena
    Gulo gulo gulo
    Hyaena hyaena hyaena
    Nasua nasua nasua
    Belone belone belone
    Lerwa lerwa lerwa
    Francolinus francolinus francolinus
    Coturnix coturnix coturnix
    Gallus gallus gallus
    Crossoptilon crossoptilon crossoptilon
    Zungaro zungaro zungaro
    Bubo bubo bubo
    Amazilia amazilia amazilia
    Coeligena coeligena coeligena
    Porphyrio porphyrio porphyrio
    Grus grus grus
    Manacus manacus manacus
    Pipra pipra pipra
    Suiriri suiriri suiriri
    Anser anser anser
    Pauxi pauxi pauxi
    Gallinago gallinago gallinago
    Caracal caracal caracal
    Lynx lynx lynx
    Genetta genetta genetta
    Vulpes vulpes vulpes
    Lutra lutra lutra
    Martes martes martes
    Meles meles meles
    Mephitis mephitis mephitis
    Icterus icterus icterus
    Curaeus curaeus curaeus
    Melanodera melanodera melanodera
    Diuca diuca diuca
    Nyctelius nyctelius nyctelius
    Buteo buteo buteo
    Cochlearius cochlearius cochlearius
    Jacana jacana jacana
    Milvus milvus milvus
    Chinchilla chinchilla chinchilla
    Tibicen tibicen tibicen
    Avicularia avicularia avicularia
    Alces alces alces
    Capreolus capreolus capreolus
    Dama dama dama
    Gazella gazella gazella
    Oreotragus oreotragus oreotragus
    Rupicapra rupicapra rupicapra
    Redunca redunca redunca
    Quelea quelea quelea
    Regulus regulus regulus
    Tchagra tchagra tchagra
    Phoenicurus phoenicurus phoenicurus
    Petronia petronia petronia
    Cinclus cinclus cinclus
    Pyrrhocorax pyrrhocorax pyrrhocorax
    Amandava amandava amandava
    Petaurista petaurista petaurista
    Marmota marmota marmota
    Hoolock hoolock hoolock
    Gorilla gorilla gorilla
    Indri indri indri
    Pithecia pithecia pithecia
    Saccolaimus saccolaimus saccolaimus
    Molossus molossus molossus
    Pipistrellus pipistrellus pipistrellus
    Myotis myotis myotis
    Chitra chitra chitra
    Microchaetus microchaetus microchaetus
    Chloris chloris chloris
    Giraffa giraffa giraffa
    Basiliscus basiliscus basiliscus
    Cerastes cerastes cerastes
    Natrix natrix natrix
    Astacus astacus astacus
    Colius colius colius
    Chalcides chalcides chalcides
    Scincus scincus scincus
    Schoeniclus schoeniclus schoeniclus
    Radjah radjah radjah
    Himantopus himantopus himantopus
    Antigone antigone antigone
    Pyrope pyrope pyrope
    Curruca curruca curruca
    Bufo bufo bufo
    Vicugna vicugna vicugna
    Erythrogenys erythrogenys erythrogenys
    $

Not that I have anything against local use---I do my stuff locally---but people running local models will very probably greatly increase demand for memory relative to cloud-based AI compute (and it's why I think that in the near term, most AI compute stuff is gonna be in the cloud, because we don't have the memory to do everything locally).

If you run in the cloud, the hardware is shared. When one person isn't using it, another can be. That can lead to high rates of capacity utilization, approaching 100%.

If I have local AI compute hardware, if I'm not using it, it's idle. If 1% of the time, I have it crunching something for me, then I'm only getting 1% capacity utilization of that hardware. That means that to provide the same level of compute capability to everyone for local use, I need 100 times as much hardware.

AI cloud companies have purchased more memory than the rest of the world is buying.

If we did everything locally, we'd need a hundred times more memory than what the cloud AI companies are picking up.

We couldn't do that without far more memory production capacity. Even if someone started on a buildout of that scale today, a new memory factory takes 4--5 years to get into production.

I mean, owls typically eat small birds too, just like cats.

I mean, you're talking about Reddit, which is a brand, in this comment.

Like, people talk about things, and a lot of things have brands.

Gorgon Halo, IIRC, goes up to a unified 192GB.

And while it depends on application, I generally agree that amount of memory is the most important factor. I started out with a 24GB RX 7950 XTX and then picked up a 128GB Framework Desktop. The larger amount of memory on the Framework is just a lot more useful than the greater bandwidth on the 7950, gives a lot more flexibility. I was always able to find useful things to do with more memory and could use more. For LLMs, more context, larger models, less quantitization. For image diffusion models, larger models, higher native resolutions without tradeoffs like having an upscaling pass, batch passes.

You can see why the cloud AI companies are hell-bent on getting all the memory that they can get their paws on.

When there is a split, the Vice President breaks the tie.

But more generally, that's an intended design feature. The Senate (power is a function of the number of states) is where the states are represented, the House of Representatives (power is a function of number of people) where the people are.

Switzerland is another federation that has a bicameral structure with a legislative house that represents the states.

Australia's a federation that has has both legislative houses allocated based on number of states.

When I went through school, I did both at various points. Sometimes I'd buy lunch in the cafeteria. Sometimes I'd take a bag lunch.

The cafeteria can do hot food, which you can't bring. But a bag lunch lets you choose options that aren't served in the cafeteria.

I mean, honestly, same subsequent to education, at work. People might bring a lunch, or might eat commercially-prepared food. Not really a universal standard.

You're not wrong, but honestly, I suspect that some Goo-Gone or similar would knock out leftover adhesive.

I have seen people recommend that shiny aluminum-backed tape as being very opaque. Dunno what the adhesive is like on that.

M-Discs are also useful as very long term archival storage.

https://en.wikipedia.org/wiki/M-DISC

In 2022, the NIST Interagency Report NIST IR 8387[25] listed the M-Disc as an acceptable archival format rated for 100+ years, citing the aforementioned 2009 and 2012 tests by the US Department of Defense and French National Laboratory of Metrology and Testing as sources.

Sure. Everyone would prefer to have a monopoly that they control, given the option (though to avoid monopolies everywhere other than that). One generally doesn't get the option to exclude other parties from competing with themselves, though.

If Delta or any other airline had the option, they'd rather be the only airline. But...here we are.

Not really a WWI expert, but I believe that you are correct that the big factor was Germany running out of manpower.

There were also technological advances, like the rolling barrage that moved immediately in front of an advance to suppress defenders. Or the early tank. But I don't think that these fundamentally changed the situation. They alome wouldn't have broken a static situation.

searches

Wikipedia apparently calls it a "creeping barrage":

https://en.wikipedia.org/wiki/Barrage_(artillery)

A report said "Experience has shown that it is far better to risk a few casualties from an occasional short round from our own artillery than to suffer the many casualties which occur when the bombardment is not closely followed up".[21] A creeping barrage was essential to the success of the Canadian Expeditionary Force in the capture of Vimy Ridge in April, 1917. The creeping barrage was used during the action of Tell 'Asur on 12 March 1918 in the Sinai and Palestine Campaign.[22] Six months later, it was used with devastating effect during the Battle of Megiddo (1918) when 18-pdr and Royal Horse Artillery formed a creeping barrage which fired in front of the advancing infantry up to their extreme range while 4.5-inch howitzers fired beyond the barrage, while heavy artillery were employed in counter battery work.[23] The creeping barrage moved at a rate of between 50 yards (46 m), 75 yards (69 m) and 100 yards (91 m) per minute.[24]

https://en.wikipedia.org/wiki/Tanks_in_World_War_I

The development of tanks in World War I was a response to the stalemate that developed on the Western Front. Although vehicles that incorporated the basic principles of the tank (armour, firepower, and all-terrain mobility) had been projected in the decade or so before the War, it was the alarmingly heavy casualties of the start of its trench warfare that stimulated development.[1][2] Research took place in both Great Britain and France, with Germany only belatedly following the Allies' lead.

Fuller's Plan 1919, involving massive use of tanks for an offensive, was never used because the blockade of Germany and the entry of the US brought an end to the war.

I think that someone could argue that tanks were becoming a more important factor in late WWI facilitating the offensive, but they weren't yet a major factor.

You could argue that air power, which was also emerging, would help the offensive, but that was even further out from becoming a major factor.

I think that it's fair to say that the machine gun, barbed wire, and the trench (maybe also the land mine, though I don't recall seeing that normally listed with the above three) favored the defensive, and that over time, other factors would come to the fore that would favor the offensive. You could say that those factors were becoming more prominent in late WWI. I don't think that one could say that those were the dominant factors in ending the war, though.

That is to say, the only defense against the new weaponry was to dig trenches, and the only way to capture trenches was to throw a few thousand men at them, at which point, you may have gained a few hundred yards.

The thing about trench warfare that makes it potent as a defensive measure isn't the idea that no trenches can ever be captured. It's that it slows and limits any offensive to the point that the other side cannot perform a "breakthrough", to both penetrate enemy lines and then be able to greatly exploit that situation. https://en.wikipedia.org/wiki/Breakthrough_(military)

A breakthrough occurs when an offensive force has broken or penetrated an opponent's defensive line, and rapidly exploits the gap.[1]
Usually, large force is employed on a relatively small portion of the front to achieve this. While the line may have held for a long while prior to the breakthrough, the breakthrough happens suddenly when the pressure on the defender causes him to "snap".
As the first defensive unit breaks, the adjacent units suffer adverse results from this (spreading panic, additional defensive angles, threat to supply lines). Since they were already pressured, this leads them to "snap" as well, causing a domino-like collapse of the defensive system. The defensive force thus evaporates at the breakthrough point, letting the attacker to rapidly move troops into the gap, exploiting the breakthrough in width (by attacking enemy units at the edge of the breakthrough, so widening it), in depth (advancing into enemy territory towards strategic objectives), or a combination of both.

Germany achieved a breakthrough against French forces in WWII---they penetrated French lines and were able to, from that point, conduct a sustained offensive that led to major gains, wirh France unable to stabilize the front.

That's the kind of thing you'd need to be able to consistently do against trench fortifications for the trenches not to be useful. If a defender can slow an advance enough that they can just dig more trenches or limit an attackers gains to a small amount and then quickly counterattack and retake them, the trenches retain their value.

that credits light machine gun units (with ample supplies of grenades) for breaking the stalemate.

I think that that tactic of having more firepower from light machine guns was tried. I don't think that it could be said to have defeated the use of trench fortifications.

https://en.wikipedia.org/wiki/Marching_fire

Marching fire, also known as walking fire, is a military tactic—a form of suppressive fire used during an infantry assault or combined arms assault. Advancing units fire their weapons without stopping to aim, in an attempt to pin down enemy defenders. Marching fire usually ends with an infantry charge to engage the enemy in close combat. The tactic requires ample ammunition and rapid-fire weapons.

The modern form of marching fire evolved in the early 20th century from a French Army infantry assault concept which suggested the use of suppressive fire from a light machine gun carried by one man—the Chauchat automatic rifle. The tactic was employed to a limited degree in World War I then further codified with the introduction of the M1918 Browning Automatic Rifle and described in U.S. small unit infantry tactics manuals in the early 1920s.

In 1903, French military theorists noticed that the machine guns of the day, heavy and relatively immobile, were of little use in infantry assaults. They determined that "the machine gun must learn to walk".[3] A marching fire tactic (French: feu de marche) was expected to involve incidental suppressive fire, with the advancing troops considered a deadlier threat than the unaimed bullets, causing the enemy to fall back.

The problem here is that it's also going to make your pricing uncompetitive.

You can convert consumer surplus to supplier surplus by using price discrimination.

https://en.wikipedia.org/wiki/Economic_surplus

Consumer surplus, or consumers' surplus, is the monetary gain obtained by consumers because they are able to purchase a product for a price that is less than the highest price that they would be willing to pay.
Producer surplus, or producers' surplus, is the amount that producers benefit by selling at a market price that is higher than the least that they would be willing to sell for; this is roughly equal to profit (since producers are not normally willing to sell at a loss and are normally indifferent to selling at a break-even price).[1][2]

https://en.wikipedia.org/wiki/Price_discrimination

Price discrimination, known also by several other names, is a microeconomic pricing strategy whereby identical or largely similar goods or services are sold at different prices by the same provider to different buyers, based on which market segment they are perceived to be part of.[1][2][3] Price discrimination is distinguished from product differentiation by the difference in production cost for the differently priced products involved in the latter strategy.[3] Price discrimination essentially relies on the variation in customers' willingness to pay[2][3][4] and in the elasticity of their demand. For price discrimination to succeed, a seller must have market power, such as a dominant market share, product uniqueness, sole pricing power, etc.[5]

But in a competitive market, normally, that won't work, because you require monopoly power to make it work.

https://en.wikipedia.org/wiki/Monopoly

A monopoly (from Greek μόνος, mónos, 'single, alone' and πωλεῖν, pōleîn, 'to sell') is a market in which one person or company is the only supplier of a particular good or service.[1] A monopoly is characterized by a lack of economic competition to produce a particular thing, a lack of viable substitute goods, and the possibility of a high monopoly price well above the seller's marginal cost that leads to a high monopoly profit.[2]

Price discrimination: A monopolist can change the price or quantity of the product. They sell higher quantities at a lower price in a very elastic market, and sell lower quantities at a higher price in a less elastic market.

And the airline industry---at least in the US, and not saying that that is the case everywhere in the world---is a competitive one.

Warren Buffet has been a pretty prominent voice critical of airlines in the US as an industry; they aren't very profitable.

searches

https://jimmysjournal.substack.com/p/why-buffett-hates-airlines-a-case

For decades, Warren Buffett’s distaste for airline stocks has puzzled many investors.
How could a business so essential to global commerce, tourism, and modern life be so unattractive to one of the world’s most celebrated capital allocators?

The airline industry is, unfortunately, a textbook example of a structurally unattractive sector:
Rivalry Among Existing Competitors: Intense. Price wars are common. Differentiation is minimal. Customers view seats as commodities, and loyalty programs offer diminishing returns.
Threat of New Entrants: Moderate to high. While aircraft procurement and regulatory hurdles present barriers, the proliferation of low-cost carriers has shown that entry is possible - often with disruptive effects.
Bargaining Power of Suppliers: High. Aircraft and engine manufacturers (Airbus, Boeing, GE, Rolls-Royce) operate as oligopolies. Airports and labor unions also exert significant leverage.
Bargaining Power of Buyers: High. With near-zero switching costs, passengers chase the lowest fare. Online aggregators have further commoditized the booking process.
Threat of Substitutes: Medium. Alternatives like high-speed rail (Europe, China) or videoconferencing (post-COVID) are becoming more viable.
Each of these forces squeezes margins. Together, they form a gauntlet that makes sustainable value creation exceedingly rare.

The airline industry has a long and notorious history of earning ROICs at or below WACC.
In other words, despite growing revenues and essential services, the industry has historically destroyed shareholder value. From 1963 to 2023, U.S. airlines rarely posted positive ROIC-WACC spreads. This isn’t just a cyclical issue - it’s structural.

Buffett never bet against airplanes - he bet against the economics of the airline business.
And over the long run, he was right.

Delta might say "okay, Person X has the ability and willingness to pay more for a product that I offer, so I'll charge more".

The problem is that that makes it very easy for the airline down the street to take their business by simply not charging that much.

And while there is some overhead associated with comparing prices, and customers will place some value on the time they spend comparing prices, there are a number of well-established existing services to do price comparisons between airlines, like Travelocity and Orbitz, which make this pretty quick and easy.

Like, I don't think that the limiting factor on an airline doing personalized pricing is likely a lack of having a model that can sufficiently-accurately predict the price an individual is willing to pay, which is what AI or other computer modeling would provide. Rather, it's the presence of a competitive market.

The destruction of the book is done to mitigate some copyright concerns. There will be a fairly small number of large-scale projects like this, and each only needs one copy of a given book, so outside of maybe very rare books, the impact is going to be pretty negligible on availability --- like someone buying one copy of that book.

I finally started putting white labels over a bunch of these the other day. It had just gotten out of hand. A Minisforum mini-PC with a blindingly bright power light was the breaking point --- that took three layers of label to bring the brightness down to a reasonable level.

I've also started actively trying to get devices that don't have power or status lights, where possible. I don't really need a USB hub with status lights. Used to be just PC components with unnecessary RGB bits that I avoided.