BANNED USER: ban evasion
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
No bio...
User ID: 3097
Banned by: @Amadan
I linked this blog post in a reply at the bottom of a long comment chain, but it occurs to me that it is probably worth discussing in it's own right.
According to all known laws of physics and aviation there is no way that a bumble bee ought to be able to fly. The bee, of course knows nothing of this and insists on flying anyways.
Wikipedia has an entry dedicated to the phrase “Thank God for Mississippi” because for the last 100 years or so, no matter how bad off your state may be in a particular way, you could typically take solace in the idea that Mississippi had it worse. "Yes, our health outcomes suck..." the the people in Wyoming and Alaska may tell themselves "...but at least we aren't Mississippi".
In my experiance shitting on the South Eastern US as an embarassing, degenerate, cultural backwater, is not only tolerated in blue and grey tribe spaces but venerated and encouraged. Of course the south sucks, that's where Mississippi is. If you are from that region and you are persuing a degree at a school like Stanford or Cal-Tech you quickly learn to hide your accent and claim to be from somewhere else if you want to be taken seriously and graded honestly by your professors.
I present this as context for...
The "Missisippi Miracle"
In 2002 the second Bush administration signed the No Child Left Behind Act into law. Educational standards and reform had been had been a big part of his 2000 campaign platform, his wife Laura being a grade-school teacher, and one of the provisions of this act was a a mandate that "Public" (that is tax-payer-funded) schools would participate in the National Assessment of Educational Progress (NAEP) originally established by the Johnson administration in 1964. As a result we now have standarized test data for almost every state and municpiple school district in the country going back over two decades.
For those outside the US, US school system is typically broken into 3 4 year long blocks. Kindergarten/Elementry School, Middle/Secondary School, and then High School. Specific names and implimentations vary from state to state but as a general rule the idea is that a child will enter the public school system at the age of 5 or 6 and graduate at the age of 18. The NAEP tests students for reading and mathematical proficiency at grades 4 and 8, IE upon entering and exiting Secondary/Middle School.
In 2003 Missisippi 4th graders where ranked near to last in the nation for reading comprehension, with an unadjusted average of 203. Only DC and Puerto Rico ranked lower. As of 2024 thier score is 219, representing a lttle over a standard deviation of improvement and placing them just shy of the top 10. This on it's own would represent admirable progress, but where things start to become unhinged is when you look at the "adjusted" figures. NAEP and various outside NGOs apply various adgustments to the raw scores in an attempt to control for things like demographics, socio-economic status, and spending per-student. When these "adjustments" are applied, Mississippi schools are not just performing better than they were 20 years ago, they are performing better than any other state school sytem in the nation. This is the alleged "Miracle".
Now a number of liberal commentators ranging from Friedliche DeBoer (of the South African Boers perhaps?) and Kevin Drum to Steve Sailer and the LA Times have all tried to debunk the so-called "Mississippi Miracle". The arguments generally fall into three broad categories. The first is that the mainstream media, academia, and establishment politicians are all prejudiced against liberal coastal blue-coded states like New York, Massachusetts, California, and Oregon, in favor of southern states like Mississippi. I find this claim laughable on it's face for reasons stated in the opening of this post. The second is the significantly more defensible claim that the NEAP's "adjusted" scores do not accurately reflect ground level truth. I believe that this is a fair critique, but the people making this critique often explicitly refuse to acknowledge that the unadjusted scores also saw an marked improvement (casts side-eye at Sailer and DeBoer) and that even when comparing like to like, the average Black student in Mississippi reads at a level about 1.5 grade levels higher than the average Black student in democratic strongholds like Illinois or Wisconsin.
Finally there is the claim that Mississippi is effectively "gaming the system". In 2013 the Mississippi State Legislature enacted the Literacy Based Promotion Act (LBPA) which required kids to pass a reading test to be promoted from elementary to middle school or else be held back or forced to repeat a year. The argument as it is, is that 4th graders in Mississippi are actually 5th or 6th graders by any other state's reckoning. If that were true one would expect to see a substantial age difference in the class cohorts, however that is not what we see, the average age of a 4th grader in Mississippi is only 0.01 years (or just under 4 days) above the national average.
To all appearances, and against the most ardent protestations of our resident Boer it would seem that having standards and enforcing them may actually matter.
How is this possible
I have a cynical answer that I expect to get me in trouble with the moderators, because I am about to take a stand in defense of Bulverism. Ad Hominem may be a formal fallacy, but in the real world it provides real value. Whether or not someone has an ulterior agenda is absolutely something you should be thinking about when you are trying to decide whether or not you are going to believe them.
I expect to be accused of "lacking charity" but the words are going to be theirs not mine. At some point all the experts in the blue and gray tribes seem to have decided that teaching kids to read was too much trouble and that not teaching them to read would be just as effective at promoting literacy as not doing so because demographics matter more than basic competency or engagement. Why would they do that even as they admitted that “For seven years in a row, Oakland was the fastest-gaining urban district in California for reading,”. The answer is in the following line "And we hated it."
By claiming that standards matter i am effectively take taking a shit on the foundational beliefs of Steve Sailer, Friedliche DeBoer, and a number of users here including at least one moderator.
Mississippi accepts your hate and Volleys it back. Ideocracy may be coming for America, but its coming for you, the blue tribe, not for MAGA country. We will teach our children Shakespeare Kipling and Twain, and you will not, and in 20 years we will see who has come out on top.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
1d ago
Section 2.1 on architecture and methodology.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
2d ago
Because I read DeepSeek's 2024 paper.
this is distinct from intelligence
I disagree, it is not distinct, it is integeral. What is the value that a human equipment operator (or any other "intelligence" for that matter) adds over a machine carrying out a set of scripted movements if not the ability to react and adapt on the fly?
Tell me do you want your "self driving car" to plow into the back of a stopped vehicle because it was programmed to drive south on [route x] at [speed] for [distance], or would you prefer that it percieve and react to the obstacle by applying the brakes and/or going around? Which of those options do you think is the more "intelligent" of the two?
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
2d ago
Muppet side-eye.png
Have they though?
The Christian conservative coalition is arguably more powerful today interms of ability to influence policy than they have been since the 90s. This influence largely stemming from the rise of Catholic and classical education as an alternative academic pipeline and the ability to point to the progressive excesses of the last 2 decades and say "i told you so".
As for the idocracy argument, it is not MAGA country that is becoming illiterate, it is blue-tribe strongholds like Oregon, California, and Maine.
I feel like i addressed @rae's objections about structure and LLMs just being token predictors within the body of the text itself. Eg
most publicly available "LLMs" are not just an LLM. They are an LLM plus an additional interface layer that sits between the user and the actual language model. An LLM on its own is little more than a tool that turns words into math, but you can combine it with a second algorithm to do things like take in a block of text and do some distribution analysis to compute the most probable next word...
@self_made_human disagreed with my definition of intelligence and approach to assessing it wich is interesting from a philosophical standpoint but also kind of irrelevant in practical terms. Fact is that adapability and agentic behavior are key things to consider when discussing whether a robot can replace a human worker, or if we're going to wakeup tomorrow to find out that Claude or Grok has suddenly gone "FOOM" and turned into Skynet, and i don't think it's "hamstringing" my (or anyone else's) understanding to point that out.
@daseindustries just seems to be angry that someone would break from the rationalist consensus.
Though aditedly taking the week of the 28th off to go on vacation probably dindnt help.
although in the age of mass immigration, I suppose the distinction is moot.
Well that would be the point under contention wouldn't it?
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
2d ago
Indeed, and as i argued in my on post on the subject i think this element of general-applicablity/adaptability is a key component of what most people think of as "intelligence". A book may contain knowledge, but a book is generally not seen as "intelligent" in the way that say an orangutan or a human is. I also think that recognizing this neatly explains the seeming bifurcation in opinions on AI between those in "Bouba" (ie soft/non-rigorous) disciplines and "Kiki" (ie hard) disciplines where there are clear right and wrong answers.
Giving you an answer you dislike or disagree with is not the same thing as not giving you an answer.
I argued, and I have continued to argue in this thread, that agentic behavior and general applicability are core components of what it is to be "intelligent". Yes a a pocket calculator is orders of magnitude better at arithmetic than any human and stockfish is better at chess, that doesn't make either of them "intelligences" does it?
While i agree with the overall thrust of your critique i want to harp on this bit
Will he change his mind if someone presents a chess-focused LLM with a high ELO score?
...i think that part of the problem is a wide-spread failure on the part of Freddie and the wider rationalist community to think clearly and rigorously about what "intelligence" is supposed to mean or accomplish. It is true that by restricting an LLM's training data to valid games of chess documented in the correct notation and restricting it's output to legal moves you can create an LLM that will play chess at a reasonably high level. It is also true that a LLM trained on an appreciable portion of the entire internet and with few if any restrictions on its output will be outperformed by a Chess algorithm written in the 70s. The issue is that your chess llm is not going to be a general tool that can also produce watercolor paintings or summarize a YouTube video, its going to be a chess tool and is thus evaluated within that context. If stockfish can reach a similar ELO using less compute why wouldn't you just use stockfish? One of the weird quirks of LLMs is that the more you increase the breadth of thier "knowledge"/training data the less competent they seem to become at specific tasks for a given amount of compute. This is the exact opposite of what we would expect from a thinking reasoning intelligence, and i think this points to a hole in the both the AI boosters and AI doomers reasoning where they become fixated on the I in AGI when the G is arguably the more operative component.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
I don't think 4chan "won" some much as the simply contributed to the blue and grey tribes collapse in influence which cleared the way for the Tea-Party to build a national mandate.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
Im not even convinced they lost the culture war. It's the liberals who appear to be imploding.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
This remind me of a thought i had watching the most recent superman movie. DC's Earth dodged a bullet when Kal'El was found by a conservative midwestern couple as opposed to literally anyone one else. As much as James Gunn, the people of Metropolis, and to some extent theMotte like to sneer at Smallville, Smallville is arguably the reason that Superman is a force for good instead of an existential threat.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
No need for it to be involuntary, just allow the degens to sterilize themselves.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
My take on this is that the US is somewhat unique in being a nation founded on a proposition rather than blood, soil, or some historical what-have-you. To that end, i believe it is in the US's interest as a nation to vet those it let's in on the basis of whether or not the are "on board" with that proposition.
Free speech is a human right, residence in the United States is a privilege.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
You are free to ping me if you like. You know that right?
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
3d ago
IF I am wrong as you so confidently claim I am, perhaps you should explain to @self_made_human that Claude, Gemini, Deepseek, Et All have none of the capabilities he claims they have. After all, if all the LLM is doing is predict the next most likely word, how do you get a chess engine or python script out of that? It's almost as if there must be some intermediate layer in between.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
4d ago
You're telling me that I am wrong and that I am ignorant, but what I'm describing is the core functionality of both DeepSeek and Google's flagship products. As I recall you are all in in on Deepseek. Do you actually read and understand any of the technical material they publish or the subsequent commentary there on? or are you a mere "think piece" writer?
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
16d ago
But we don't exclude them do we?
More like saying that the soyuz rocket is propelled by expanding combustion gasses only for somone to pop in and say no, its actually propelled by a mixture of kerosene and liquid oxygen. As i said in my reply below, you and @self_made_human are both talking about vector based embedding like its something that a couple guys tried in back in 2013 and nobody ever used again rather than a methodology that would go on to become a defacto standard approach across multiple applications. You're acting like if you open up the source code for a transformer you aren't going to find loads of matrix math for for doing vector transformations.
The old cliche about asking whether a submarine can swim is part of why I made a point to set out my parameters at the beginning, how about you set out yours.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
16d ago
Well said.
C'mon dude. If this is the third draft of the essay, I really expect more substantial rebuttal than this.
You misunderstand me. My response was not the third revision, it was the third attempt.
I don't know if you realize this, but you come across as extremely condescending and passive-agressive in text. It really is quite infuriating. I would sit down, start crafting a response, and as i worked through your post i would just get more angry/frustrated until getting to the point where id have to step away from the computer lest i lose my temper and say something that would get me moderated.
And that illustration was wrong.
As i acknowledged in my reply to @Amadan it would have been more accurate to say that it is part of why LLMs are bad at counting, but I am going to maintain that no, it is not "wrong". You and @rae are both talking about vector based embedding like its something that a couple guys tried in back in 2013 and nobody ever used again rather than a methodology that would go on to become a defacto standard approach across multiple applications. You're acting like if you open up the source code for a transformer you aren't going to find loads of matrix math for for doing vector transformations.
Why is the opinion of the "average American" the only standard by which to recognize AGI?
Why isn't it a valid standard? You are the one who's been accusing society of moving the goalposts on you. "the goalposts haven't actually moved" seems like a fairly reasonable rebuttal to me.
I had forgotten how much of your previous weak critique to the same evidence was based off naked credentialism. After all, you claimed:
I understand how my statements could be interpreted that way, but at the same time I am also one of the guys in my company who's been lobbying to drop degree requirements from hiring. I see myself as subscribing to the old hacker ethos of "show me the code". Its not about credentials its about whether you can produce tangible results.
The companies that spend hundreds of billions of dollars on AI are doing just fine.
For a given definition of fine, i still think OpenAI and Anthropic are grifters more than they are engineers but I guess we'll just have to see who gets there first.
TequilaMockingbird
Brown-skinned Fascist MAGA boot-licker
16d ago
As i have said in prior discussions of the topic, I fully believe that AGI is possible and even likely within my lifetime, but I am also deeply skeptical of the claims made by both AI boosters and AI doomers for the reasons stated above.
Citation needed?
Know your meme.
More options
Context Copy link