saalweachter

Oddly, there are now more Tandy Leather Company stores than stand-alone Radio Shacks in the US.

It used to be the norm for companies to become profitable, then IPO, moreso for megacorps.

If I am remembering my history correctly, the first company to go public with a multi-billion dollar valuation while being unprofitable was Netscape, which kicked off the dotcom bubble.

You just raise the head.

You harvest the top crop, the one you want, cutting above the lower height, then go back through and make a second pass with the head lower.

You'll lose some of the lower crop, crushed under the wheels, but it was there as bonus anyway.

The glacial cycle was longer duration, lower fluctuation.

Glaciers don't melt instantly the moment the temperature flips from "form glaciers" to "melt glaciers". They operate on the integral of temperature -- heat. The temperature determines how fast they melt or form.

These melted really, really fast.

It requires management buy-in, both to allow some number of engineers to be allocated to such "non-production" work, and also to allow the friction security creates to continue to exist and "slow down" the "real work", both by vetoing things that would weaken security and by forcing everyone to use more cumbersome workflows.

That usually doesn't happen unless the company has suffered a large financial or reputational loss as a result of poor security practices. Sometimes not even then, if the company doesn't get scared they'll start losing money or get in some sort of legal trouble.

I think "good security/engineering practices" still has a chance.

Things like--

  * Minimizing attack surfaces by not running lots of software on the same server and not bloating feature sets beyond the minimum 
  * Limiting access so that most things can't be done by most users, there are *no* super-users, and escalating access is temporary ans requires multiple accounts to authorize
  * Encrypting data at rest

All software has flaws and it's now easier than ever to find and exploit them, but this disproportionately hurts people who were only putting on a show of security and relying on a single layer or two of defense instead of developing a robust insider threat model (which helps with both insider threats and threats that get inside) and defense in depth.

It also doesn't necessarily protect you from anything by obtaining it.

If you, say, go and get a certificate of no-permit needed when you're replacing all of your kitchen cabinets, and then you decide to move the sink while you're at it, you're back to still needing a permit for the plumbing change (and even if you aren't, if your neighbor calls the inspector and says they think you might be, the inspector might say, "no, we already know about that, it's just cabinets", or they might wonder if you are and then you're back in the situation you'd be in without the certificate).

In the US there's also a third category besides "permit approved" and "permit denied", "no permit needed": the thing you are doing is allowed and not covered by the permitting process. You just do it. This shows up in both building code (you typically don't need a permit for most repairs or cosmetic renovations) and environmental protections (you don't need an environmental study for logging in a non-protected forest).

The tricky part with "no permit needed" is that while you can just go and do whatever, if someone charged with enforcement suspects you're doing something that requires a permit, they can still demand you prove you don't (and stop work until you do), and if you guessed wrong and you did indeed need a permit, you're double-fucked.

That's why for many of these processes, there's also something like a certificate of no-permit needed, where you go through an abbreviated permitting process spending a lot of time and money while all of the clerks and officials stare at you like you're a moron before finally giving you a piece of paper saying you didn't need to do any of that in the first place.

And that's why a lot of people with a lot of experience and a good understanding of the law skip getting a "no permit needed" certificate.

I am assuming it's similar in other countries.

The German spelling reform was also an international agreement by Germany, Austria, Lichtenstein and Switzerland.

Can you imagine trying to organize a spelling reform agreement between the US, India, Pakistan, Nigeria, Philippines, the UK, Germany and Canada?

I mean, I would sign up for a study where I am randomly assigned a hobby to pursue for the next 20 years while researchers study my health outcomes.

I think this needs a catchy name.

It's not a new problem in software, and it's not unique to software.

A lot of hobby woodworkers, for instance, spend most of their woodworking effort on their new woodshop. Building a workbench, building shelves and cabinets and special holders for all their drill bits, mortising chisels and tools, special tables for their miter saw, a special cabinet connected to their dust collection system for sanding small pieces...

Likewise, the favorite project for programmers is ... programming tools. New languages, new libraries, new editors. We've gone through like six version control systems at my day job.

There is some merit in this, since new tools make you more efficient at building more new tools, but eventually you have to step back and remind yourself what your actual goals are.

I call it the tool maker's dilemma, but that's not nearly as catchy or easy to fit into conversation as bikeshedding.

For the longest time, exchanges didn't list stock with super-voting/dual class shares. It was only in the late 80s/early 90s that this rule was dropped; you'll note that Microsoft and Apple don't use them, as they were banned when those companies went public.

You're not getting high budget shows without like a hundred million subscribers, sure, but there are bootstrapping options.

Dropout does their original content on around 1 million subscribers @ $6/mo.

There's also licensing existing content -- the mega corps building walled gardens aren't likely to play ball, but if you make it painless and have non-exploitative terms, there's countless smaller content producers who would happily upload their catalogue to another service.

I think if the right group of founders (experience in streaming,v experience in content production, good professional network with existing smaller production companies) was able to kickstart it with a million subscribers, give or take, they'd have decent odds of providing enough value to start a more subscribers -> more content spiral instead of the reverse.

Netflix isn't owned by its subscribers is the critical difference. You don't buy shares of NFLX to gain access to the service, and subscribing does not come with any shares/voting rights.

What I want is a streaming service that operates as a consumer cooperative.

Subscribers vote, elect a slate of directors who each get a fraction of the content-budget to spend as per their campaign promises. Licensing content, buying it in perpetuity, producing it themselves or funding some group of creators -- none of this "how can we milk customers more", and no more fruitless community campaigns to save a beloved series, you just need some critical mass of subscribers to vote for the guy whose platform is "I will keep making more seasons as fast as possible until the sun boils the ocean".

For a lot of environmental regulations, whether or not a permit is needed is most of the question.

In the US, if you want to build a shed in the middle of a wetland, you need to go through an arduous permitting process and are unlikely to gain approval. If you want to build it within 100 feet, you can probably do it, but you need to confer with the relevant DEC to show that you won't negatively impact the wetland during construction. If you're more than 100 feet away, you don't even need to talk to the DEC, because you're not in protected land.

Of course, if the DEC decides they don't think your shed is outside the 100 ft adjacent area, they may decide to do a demarcation and figure out exactly where your shed is relative to the wetland, and if you messed up you're in trouble. But if they send out a biologist to demarcate the wetland and you're far enough away, you're fine.

So many people, if they think it's ambiguous, will take the time to go through an abbreviated permitting process to get a certificate of "no permit needed", to save themselves the headache later on.

If you go through a process like this, chances are people with local expertise will tell you you're wasting your time -- the case looks pretty clear cut, and hardly anyone ever has trouble in such circumstances. The clerk at the department responsible for issuing the "no permit needed" certificate may tell you you're wasting everyone's time.

But if you're a big corporation like Google, you don't really have an excuse not to do it. You can afford the lawyers, you can afford to plan years in advance, and so many more people are going to notice if you skip steps. Even if you end up being in the legal clear, no permit was needed, you're so much better off getting it in writing in advance.

... you realize Asia consumes 3-4x more pork than tofu, right?

There's a handful of reasons to keep even beef in the equation, albeit at way, way lower levels than we produce it now.

One is co-production of other materials -- wool is another desirable material.

Another is optimizing land use. Some land isn't suitable for row crop agriculture, but can sustain grazing. There's also land we're irrigating for row crop agriculture right now from aquifers we really shouldn't be; converting it back to grassland and grazing on it would let us recapture some of the lost calories.

Then there's un-optimizing the food chain. From a food-per-acre standpoint we should be primarily subsisting on soy and whichever of corn, potatoes or rice grows best in the local soy & climate. Even if we are going all in on minimizing land usage, however, we don't want to produce exactly as much food as we need, because crops fail and famines are bad. You can aim to overproduce a little year to year and maintain a surplus of storage to cover the bad years, but you can't just cover the average bad year. You need to cover the worst years possible happening several years in a row, you need to be covering once-in-millennia bad luck. So you need to overproduce enough that you don't expect some of it to ever be consumed by humans.

So you need something to do with the extra corn you produce, besides feed it to people, that you can rapidly cut back on in case if crop failures to return food back to the supply. One option is ethanol, and another is livestock -- feed your surplus grains to livestock, and then eat them instead. If you have crop failures, you cull the herds, which frees up their feed stocks for human consumption, as well as adding the extra meat calories to the supply. (And this even happens somewhat automatically with market forces -- farmers are always arbitraging grain prices against meat prices, and a spike in grain prices will lead them to sell off livestock as they become less profitable, which will in turn depress meat prices further.)

If a tank of 10k PSI compressed air ruptures, you're going to know it.

I am going to drive my Chevy Volt into the ground.

The point of sending them off to a carefully selected globular cluster would already be to give them a better life.

I'm not saying someone won't convince everyone on board that a silver future for their Nth-great-grandchildren worth derailing the golden future of their 10Nth-great-grandchildren (especially if it looks like the generation ship won't last 10N more generations), but from my POV sitting on a couch that isn't on a generation ship (well, not something traditionally defined as a generation ship), it doesn't feel like something you'd get everyone all fired up over.

Yeah, but why. It's not like you could redirect it enough that you'd land somewhere in your lifetime. It'd just be changing where the ship ended up in 10, or 100, or 1000 generations.

Re: changing course, as long as your generation ship accelerates for more than one human lifespan, your future generations of involuntary colonists aren't going to have a lot of options besides going along with it. The choice wouldn't so much be "keep going or veer off and colonize a closer star system" as it would be "keep going for another 800 generations or veer off and colonize a closer star system in another 80 generations". I mean, there's probably no way to stop them from making that choice, but it's hard to see anyone caring whether their nth-great-grandchildren die with one or ten millennia left on the trip.

That's why we need to find a globular cluster with metal-rich planets, and take a generation ship over there.

Sure, a thousand generations will live and die along the way, but once we get over there, you've got stars every 0.1 ly or so, and with very plausible future tech, you could navigate between them in a few years. You could set up a proper starfaring, interplanetary civilization, with tradeships either going back and forth between two neighboring systems, or winding on long, never-repeating routes, visiting a dozen or so star systems in a human life.

(Most globular clusters appear to be old and metal-poor, and we haven't found a lot of planets there, and this only works if you find a globular cluster where ~every star can support a colony.)

You've got two huge filters on C++ codebase quality.

First, you have the "our core libraries/dependencies were written when it was 'C with Classes' and we're going to keep using that style for consistency/laziness/because it's well-tested", and then you have the "C++ is ugly/full of segfaults/hard, I'm going to use <new language> which solves all these problems".

You have to pass those two filters before you get to people who are interested in writing good, modern C++.

Ukraine actually did finally ratify the Rome Statute; we're down to 29 signatories who haven't ratified it.

Random questions--

Were you satisfied with the paper?

Having read the paper, do you understand "what you missed" in those 3000 hours?

From the perspective of "is this just a garbage dump to impress the market?", it's way too easy for investors to figure out 700 films are garbage.

Wikipedia just told me there's a galactic algorithm for integer multiplication in O(n log n) based on FFT so I'm guessing those two proofs are related.

My gut is that you could distill 98% of what's currently being done inside LLMs into a classical knowledge base/inference engine and only use the language models for natural language processing and an oracle for brainstorming and cut most of the computational cost and hallucinations out.