I can be reached at hnthwarted (&) thwartedefforts (x) org
https://github.com/thwarted
#D5B390 #274e69 #FF9933 #7a8c96 #663399
- PJ O'Rourke had a line in his book "Parliament of Whores" when he, as a layman, ham-fistedly cuts a bunch of stuff from the federal budget, and then just subtracts 10% from it at the end. Probably not the originator, but a quote I think about often.
"Add it all together, and I've cut $282.8 billion, leaving a federal budget of $950.5 billion, to which I apply O'Rourke's Circumcision Precept: You can take 10 percent off the top of anything. This gives me another $95 billion in cuts for a grand total of $337.8 billion in budget liposuction."
Parliament of Whores, page 103.
- Except when it doesn't because of browser or platform differences/incompatibilities.
- A Google search for vitamin d results in ads, ahem "sponsored results", for 180 servings for $27, which is about $55 for a full year assuming it's one serving per day, which is the same decimal order of magnitude as $10 (but, I suppose, since we are on HN, is three or four orders of magnitude in binary)
- > It won't scan cards with AI - you manually enter the barcode, which I think makes it less prone to error.
This is a very interesting sentence.
I interpret this sentence as saying that manually entering a barcode is less error prone than letting AI do it, that AI would have an unacceptable margin of error (and this is probably an accurate assessment).
But you don't need AI to find or read barcodes. Finding and reading barcodes is a reasonably mature technology that has existed long before AI.
Barcodes exist as a fast, machine readable data transfer format meant to avoid data entry errors by avoiding manual data entry, and yet you've implemented manual entry in order to avoid errors?
Now, if one of the constraints you've put on your implementation is that it work only in the browser and you don't want to have to download a large barcode scanning library to the browser, then it makes sense to implement manual entry. But that has nothing to do with AI.
That being said, there are some barcode reading apps that can be used to prompt for a scan from a web page, and you get the barcode payload back. I've used an app called "bineye" on Android (source on GitHub) that works like this. This helps avoid error prone manual entry and gets the full barcode payload (many barcodes store/encode more information than the human readable text printed next to them).
- What's missing in this writeup is if/how capsudod can be configured to drop/acquire capabilities (in the capabilities(7) sense, in a namespace sense, or using openbsd's pledge(2)) before executing the command. From looking at the code it seems to rely on filesystem permissions to access the capsudod socket; when in order to make mounting USB drives actually safe via this kind of permission proxying method, you'd want to only give it very limited access to operations, like only the ability to make the mount(2) syscall. I generally agree, this is sudo in a different format and with a daemon; perhaps that's better. And it's written in C, which doesn't speak to avoiding the classic C-related privilege escalation risks (although this is a small amount of code compared to sudo).
- And there's another group, grifters, who are neither living to work nor working to live. They are the parasites, and our current society rewards grifters by not putting them in check. Probably because so many want a piece of the grifting pie, in the same way many people see themselves as temporarily embarrassed millionaires.
- You need conditionals and loops. Recursion counts as looping.
- > Or you have a leader above who has no idea and goes with the quickest/cheapest option.
This leader is not going with the quickest or cheapest option. Doing so would probably be laudable. They are going with the claims made by someone that a certain way is going to be quicker or cheaper. It doesn't matter if it actually is, or ends up being, quicker or cheaper. One plan is classified as meeting the requirements while another plan is classified as being cheaper, the cheaper one will be chosen even though it doesn't meet the requirements.
- We can blame von Neumann (et al) and his infernal architecture, where memory stores both instructions and data.
- Sometimes achievements speak for themselves and provide the marketing for the actor. But that requires both the achievement to be extremely outsized, so as not to get lost in the noise, and very obviously the result of a singular actor. Only one person can step up to the plate and swing the bat.
- > When I open a PR without discussing it at all beforehand with anyone, I expect the default to be that it gets rejected.
TNG S2E8, "A Matter Of Honor" is about this topic. The submitter introduced risk on the maintainers (the risk being here largely eating up the maintainers time needlessly) by working in isolation and only presenting the finished work without any feedback or awareness from the rest of the participants.
- The compose key on Linux makes deliberate use much easier (rather than automatic replacement which often triggers when I don't want it). There's a compose key utility for Windows, but has some minor annoyances like many input (mouse or keyboard) macro extender applications.
- Emdashes are useful for an embedded appositive phrase, which a colon can't handle the same way.
- That's almost more damning. The list was created by humans, who presumably read the books, but then couldn't be bothered to summarize the very books they read? Either the human is really lazy ("read" the book but can't be bothered to write a short summary) or really really lazy (didn't read the book but felt a summary was necessary). Either way, it makes this list less interesting, at the very least because it doesn't need to exist at all if someone can just ask an LLM "list and describe books that A16Z might think are valuable to read" and get the same quality output.
- "Automating agency" it's such a good way to describe what's happening. In the context of your last paragraph, if they succeed in creating AGI, they won't be able to exercise control over a robot army, because the robot army will have as much agency as humans do. So they will have created the very situation they currently find themselves in. Sans an economy.
- > they never go into buildings, the buildings all have similar size, the towns have similar layouts, there’s numerous visual inconsistencies, and the towns don’t really make sense
These AI generated towns sure do seem to have strict building and civic codes. Everything on a grid, height limits, equal spacing between all buildings. The local historical society really has a tight grip on neighborhood character.
From the article:
> It would also be sound, with different areas connected in such a way to allow characters to roam freely without getting stuck.
Very unrealistic.
One of the interesting things about mostly-open world game environments, like GTA or Cyberpunk, is the "designed" messiness and the limits that result in dead ends. You poke at someplace and end up at a locked door (a texture that looks like a door but you can't interact with) that says there's absolutely nothing interesting beyond where you're at. No chance to get stuck in a dead end is boring; when every path leads to something interesting, there's no "exploration".
- Presumably, "design" here is referring to the visual, graphic design.
> It is pretty interesting because it is true that design is better solo'd than committee'd.
This applies to all manner of "design", especially when it's at the edge/interface, which includes humans interfacing with a tool, but also with interfaces such as APIs. It's hard to maintain consistency and coherency and vision in the design of anything when it's committee'd and/or a free-for-all.
O'Rourke's take is an interesting read; it is commentary that is meant to be more humorous and entertaining than political, I think he excelled at that in the entirety of Parliament of Whores. It was published in 1991 in a different political climate. He does admit he's doing this for fun, that the takes he express are mostly uninformed about the nature of many of these government departments and programs, and takes a (traditional) conservative (high level, and ahem, naive) view of many government programs. For example, additional quotes from that PoW chapter:
> Training and employment is properly the concern of trainees and employers: $5.7 billion.
> Insurance companies should gladly pay for consumer and occupational health and safety: $1.5 billion.
> If unemployment insurance is really insurance, it ought to at least break even: $18.6 billion.
I shared this for the Circumcision Precept bit; the portions of the quote surrounding that were context.