HermeneuticsOfVibes
No bio...
User ID: 3708
Makes sense, I think it all comes down to the corpus of public information available. If similar work can be scraped from one of the half billion repos on github its probably great, otherwise it likely isn't. Most of my work is either web stuff, infra as code, or business logic governed by public policy (rather than internal rules) so it's pretty ideal.
I imagine it would be significantly worse if I were still in global finance, but we often had the same issue as the LLMs. Lack of information and often having to switch tasks while we wait a week for a request for some bespoke implementation details to go up the chain, over to IBM, and back.
They are undoubtably useful, especially in a domain like writing tests. If you are writing tests post-facto, you have a very well defined problem with a constrained domain, and a massive corpus to consult in terms of style, strategy, etc, etc. We live in a world where hundreds of millions (and I'm probably lowballing) of tests have been written, including tests for some of the hardest math and logic problems known to man. Every leet code website and programming competition out there uses a suite of test cases to verify correctness of submissions.
Honestly writing software tests is almost the optimal use case for an LLM.
I completely understand your experience regarding the Russians. In any given domestic situation, the same character is given four different names, and none of these are what his coworkers call him.
Honestly I think the price of ec2 in terms of server time is somewhat reasonable. Not reasonable-reasonable, but like within 4x the cost of actual hardware and electricity, and honestly it's close-to-cost if you sign a year long contract for provisioning.
The way they try to fuck you on bandwidth though is just beyond the pale.
Based on a bit of research I'll probably just slop this out with Hetzner cloud. Their stuff lives in Germany, they charge $.0012/gb going both ways, and ultimately all I need is a dumb pipe that I can kill with little commitment.
Been meaning to build my own VPN again for a while, had one many moons ago but it died with the ancient blade server a got for free a decade ago.
Trying to think through how to architect it, since I want something off-prem, preferably in a different country. I could just do it how I'd do at work and have something in like an hour, but then Amazon would rob me to the tune of $.09 per gigabyte. Digital Ocean would instead want $.01 per gigabyte, plus a flat $5 or whatever, but it still feels like I'm paying through the nose for bandwidth, and the markup is still crazy.
How does Amazon get away with charging like 1200x the commodity price of bandwidth of a data center provider like hurricane electric?
I use the tools quite a bit, especially in the use case of architecting cloud infrastructure, where as you might imagine, every solution you are building has already been done by 100,000 other people before you.
In general there's something to be said for point three, but not how the speaker intended it. If an LLM is incredibly helpful and can eliminate 50+% of your work, most of your work is repetitive junk that can be automated. Just like traditional software in the 90's and 00's automated away a tremendous amount of human labor that amounted to manual data entry, LLMs are very good at automating tasks that are the next level above that.
What do I mean by the next level above manual data entry? Tasks that are performed by many, many people, in only slightly different ways, across many different organizations. Let's say you work in HR and have to compile a weekly email for exec leadership that serves as an analysis of discretionary expenditures, based on some grouping, where the data comes from some attached spreadsheet. This is an incredibly common task people are performing, with many extremely common sub patterns (i.e. huge spend by R&D this week on flights and accommodations, they are attending a conference and chose to file it as discretionary instead of getting training and travel pre-approved). Because there are only so many unique scenarios, LLM's are a good use case for automating this kind of work, even if it is a dynamic analysis task.
The reason engineers and scientists find less general utility from using LLMs is likely just because they aren't walking well trodden ground. Were I in one of these roles that benefit so highly, I personally wouldn't be bragging about how much of my workload is able to be automated.
/pol/ isn't a self hosted 4chan offshoot, it's just 4chan. And the correct comparator to reddit isn't even 4chan, it's twitter, where people are happy to issue calls for the death of whoever you'd like, unabated.
Memory is such a weird, non intuitive thing. Something can live inside your memory for eternity, for no reason whatsoever, but if everything did that, we would not even be able to function. Why is it that I will forever know the name of, and be able to pronounce, the state fish of Hawai'i, but I can't recall my mother's middle name?
(I'll save you the google search, it's the humuhumunukunukuapuaa)
I'm curious as to what specific media you are referencing. From my own small, myopic perch, 80's media seems to be all over the place in terms of future predictions, with the major constants being:
- They like to make media about the future in the first place, and
- They foresaw an increased ingression of corporate presence in a person's daily life
The latter was almost certainly because they themselves were living through a period of rapid corporate expansion into the personal domain that probably seemed like an endless march. That said though, I'm not super well versed in 80's media, and what little I have seen or read has moral lessons and warnings all over the spectrum.
I strongly disagree with laying blame on the Catholics, and I'd actually lay the fault at the feet of culture of the Chesapeake pointing fingers elsewhere. While you could view the Virginia and Maryland colonies as high trust through the eyes of the planter class (in their intereactions with members of the same class), it was absolutely a low trust, chaotic mess in every other regard, and much of modern American low trust culture has its roots here.
From the everpresent threat of rape looming over every woman by men of a higher class than her, to the significantly higher crime (and especially property crime) rate compared to the other colonies, to the absolute reverence for individual freedom (including the freedom to enslave), to the near worship of fortune and luck as a prognostication of God's general favor, to the gentry asserting themselves as arbiters of what messages the clergy can deliver, to the most popular lesiure activies of all classes and ages being the slaying of some sort of animal (in porportion to their rank in society), to prohibitions on education of both the slave and servant classes, to the ubiquity of class condescension, to the general preference for violence and permanent disfigurement as means of punishment for transgressions, the entire society was structured to create about as little trust as a highly decentralized society could ever managed.
Anyone of a higher class interacting with one of a lower had to rightly worry that they were dealing with a violent savage with a short fuse that could snap at any moment. Anyone of a lower class intereacting with one of a higher had to rightly fear that they'd be subject to any and all forms of abuse with no possible form of redress.
Another driving consideration is that of all the original colonies, the bay colony, and later the south as a whole, saw the largest geographic redistribution to the greater west of any of the original colonies, and largely brought their culture with them. In most cases of 19th century inland immigrant migration, the immigrants were moving to places already well tread by Anglican diaspora, and were subject to their existing practices. As elsewhere, the first settlers have a massive, disproportunate impact on the culture well beyond their size (as @quiet_NaN also correctly points out about the Quakers, who might have the greatest impact:population ratio of any American migrant wave).
A couple of causes. The first, that ThomasdelVasto gets at, is that you as a manager have developers, you have a domain space in the business, and you need to generate an endless stream of work to justify the continued existence of your position and the positions of your direct reports. When all you own is a hammer, you are very incentivized to find an endless world of nails.
The other is that top level leadership across tech currently has no vision, and when leadership doesn't have vision they instead default to bullet point lists of "stuff". This cycle repeats itself endlessly, a company is created to solve some problem (like JIRA/Kanban apps all sucking), succeeds (like trello), loses it's divine spark (built the thing, was successful enough to get bought out by Atlassian), and succumbs to endless feature bloat just like all of its predecessors. The "stuff" lists are also how vendor software, security practices, etc are generally decided. A committee of mixed interests will always go with the vendor that meets the most bullet points on your list, regardless of the product's performance, intuitiveness, or other soft metrics that can't be meaningfully added to a checklist.
Re: The last thing, we're just two people talking, I don't see the value in calling something out as irrelevant.
Re: The first, I don't even think you can reach the general idea by way of this sort of quantitative analysis, but I do have something better (an anecdote) that your specific sub-example reminded me of:
Once worked at a very self-important, corporate place, where the break room constantly smelled like curry. It was absolutely overwhelming. One day, someone put up a passive aggressive anonymous note in all caps asking that people "PLEASE STOP MICROWAVING CURRY" because it made the whole building smell and made them unable to taste their own food, or something to that effect. Guess this landed on HR's desk pretty fast, because the next morning the entire office gets an email about racial microaggressions will not be tolerated, and now hundreds of people had to take a racial bias training course (most of whom happened to be Indian).
Superficial just means surface level or shallow so far as I typically use it. Lots of superficial attributes are important in achieving various outcomes. We haven't elected a president shorter than 6 foot since Jimmy Carter, there's about a 0.04% chance of that happening by random chance if that superficial characteristic is unimportant to electability. Google's "41 shades of blue" experiment likely cost millions in developer time for a superficial change to hyperlink colors. And so on.
There's significantly more baggage that comes with being American than just being born in a particular place. And whether or not an American wants to be an agent for the empire, they are. They spread their American ideas, American perspectives, and other products of the American culture, as effortlessly as they breathe. When most people in the English speaking world go online, they feel like they are digitally transported to America. Not only the people but the structures, the dominant ideas, etc, are all American. Americans on the other hand, almost never seem to experience this sensation.
A critical third leg of Liberalism is contractualism. There must be a set of rules that limit those with great amounts of power, such that the common people have some inalienable cloister to retreat to. Neither Assad's Syria or Somalia in the 2010's were liberal states because you had to live under the constant threat that some powerful psycho would ruin your life on a whim with no reprisal.
This leg is probably the most collapsed of the three, and generally three begets two begets one.
People spend the majority of their waking lives largely doing these things, they are important regardless of superficiality. Most people's impact on the world is primarily what they do, not what they think. And even in terms of ideology, MeanRedMan and MeanBlueGuy are most critically, promoters of the American cultural hegemony and distributors of various propaganda.
For what it's worth, the cultural divide between red and blue America is still far lesser than the divide between the two great bay colonies that would eventually unite against the crown, even if current levels of general contempt for the other are about the same.
The polities live similar lives, eat similar foods, consume similar mass media, display similar politics on issues primarily determined by age (social security, medicare), enjoy similar past times, have similar incomes, etc. The median red and blue voter are both very identifiably American when mixed into a global pool of people.
Its easy to pick out salacious examples of this not being the case, but how much of this is driven by:
a) The various outrage optimization engines, and b) One's own human tendency to remember the remarkable and aggregate exceptions, ie to over pattern match
Field and institution are everything. Are you seeking a professorship at Yale in the Women's, Gender, and Sexuality Studies department? If so you'll be waiting for a cold day in hell. Are you seeking a professorship at Colorado School of Mines in Petroleum Engineering? If so you were never at risk.
As for whether or not a position is attainable, odds are grim, for multiple reasons. The evergreen issue of academia being a pyramid scheme where successful P.I.s and labs are built on the back of the dirt cheap labor of grad students and post docs still applies, and most of those people will still never hold a professorship, and most of those professors will never receive tenure. Things are slightly worse now because of the job market (when private industry tightens purse strings staying in academia always becomes more appealing), but only by degrees.
If you want to shoot for a professorship, be prepared to work very hard to compete against other people who are among the very best in the world at the very thing you are doing. The system nearly as much rigged in favor of woke types, as it is rigged in favor of people willing to nolife grind out a strong publication history early in your career.
I largely agree with you. I think the difference is probably (and we may never know for sure) what are they optimizing for now more than how they are going about it.
I think 2015/2016 social media companies were really optimizing for maximizing the attention as their one true goal. Whereas by the time we were deep in the covid years, they were seeking to metacognitively reflect their understanding of you back to you, while continuing to optimize for attention.
This has been ongoing for far longer than that. Tristan Harris's TED talk outlining how he as a google employee explicitly aimed to manipulate you to maximize your "Time On Site", came out in 2016, and his original internal talk on the subject dates back to 2013.
A few additional data points:
- Twitter started phasing out the chronological timeline in favor of the engagement algorithm in February of 2016.
- Instagram switched from chronological feeds to "best posts first" in the summer of 2015.
- Facebook as of 06/15/15 acknowledged that news feeds action on how long a user reads a feed item, as well as how likely they are to engage with it.
Relabelling an unpopular idea rarely, if ever, works. The fundamental problem is that free speech, called by any other name, is no longer popular. If you doubt this fact, find whatever the most incinidary post of the day made by a [left/right] aligned person is, then ask a [right/left] aligned group if said speech should be "platformed".
The reality you have to operate in now is one where free speech, is deeply unpopular, insofar as said speech is something inflammatory that the reader is outraged by. The slogan you so despise is just people broadcasting their honest, genuine intent: "there should be consequences to speech I disagree with".
I don't think it's good for anyone, but the average American adult is pushing past 7 hours a day at this point. Gaming is among the least harmful applications though, given certain parameters (played with friends, in a setting where you directly impact each other cooperatively or competitively).
As loathe as I am to generally recommend an anime, Serial Experiments Lain had me milling over this question a lot as well. My main objection to it is that the people who do this tend (not always, but there is a general pattern) to become the mask rather than taking the mask off, and usually in a self-destructive way.
Who is this person, really? How much of their being is dictated by their immediate environment?
There aren't, but there are plenty of Amish a stone's throw from Cleveland, Columbus, Philly, and Pittsburgh. They even open furniture stores in these cities, and employ the English to run the stores, build websites, do customer service, and what have you. Doesn't take as much distance as you'd think.
But it is a straightforward social contagion, in the same way female tourettes is. You can track the prevalence of specific, non standard clinical symptoms that gain popularity via one or a handful of "content creators" that explode out of no where, and trivially make the case that people, overwhelmingly teenage girls, are cargo culting various mental illnesses.
- Prev
- Next
 
			
I remind my kids of this frequently, and my wife occasionally. Very important to maintain an internal locus of control, especially in such a highly externalizing society.
More options
Context Copy link