I was one of the people who built the "plugin based ranking model" at Semantic Scholar as an intern several years ago; it's a rare treat to get a blog post about what happens after you leave and they need to replace your system.
Although actually I feel validated in one respect - it sounds like it took a huge amount of data cleaning and human validation to use the click log data for training.At the time we got many questions about why we didn't "just use it". Although now that I'm at a company with several orders of magnitude more traffic I am a believer in the almighty click log.
Yeah, I think the post author may be confused. I've migrated several high traffic gRPC services (basically TensorFlow Serving) from AWS NLBs to ALBs for the improved load balancing. It worked pretty much as desired (although a little expensive).
I am annoyed by the article's use of Bill Gates to exemplify the "fortuitous circumstances that figure into every success story". Bill Gates and Microsoft is not "every success story". I would be surprised to find even a hardcore believer in an ideologized meritocracy who thought that tail events like Bill Gates and Microsoft didn't have a huge luck component.
Meritocracy-as-strong-just-world-hypothesis is a weak-man argument.
I am a believer in meritocracy. But, I note, I'm not a believer in any ideological big-M Meritocracy. Rather, I think meritocracy is the best practical way we know to organize things. In a liberal society, rather than get everybody to agree on the latent virtues they love the most, I think it makes sense to have demonstrated capability as a common ground evaluation metric. As another poster in the thread suggests, what do you want from your surgeon if not demonstrated capability?
The "tech bro" angle is just clickbait. I am pretty sure that Foxconn, Tesla's Gigafactory, and a datacenter are not what is usually meant by "tech bro".
Second, if cities believe that these tax breaks are good investments then there is little relationship between making those investments and putting money into other programs. Cities should have access to cheap-ish capital and future tax breaks seem like fairly favorable conditions. If anything, that cities need money is an argument for investing in business (if you accept that doing so makes money).
The article doesn't much of an argument that these investments are bad. It just kind of waves in the direction of some dots and suggests you imagine some connections. If you look at the linked article about Missouri budget shortfalls you won't find "Cerner" anywhere - for all we know that was a great investment but other bad bets were made.
It is true that an algorithm can be converted to a formula, and a formula to an algorithm, but it is unlikely a given useful algorithm could be generated mechanically from a (human-written, useful) formula.
For example, you could write a formula to express the property of order in lists, and maybe with some mechanical procedure (falling under "obvious to the person skilled in the art") you could then generate an algorithm that sorted lists. However, what's the time/space complexity of that algorithm? What are the practical runtime characteristics? For industry applications the practical runtime characteristics matter as much as anything else. That you technically could compute something doesn't matter at all, if that computation might not finish until relevant celestial bodies have phase changes.
I'm not saying Quicksort should be patentable, but I do think it's at least a step and a half removed from pure math.
There exist counterexamples in the broader world. Historically in the Sinosphere it was commonish that two people might share a command of written Classical Chinese but not really be able to speak to each other. For a modern example, consider the paper below:
https://eric.ed.gov/?id=ED515291
"human convention of written language" is a bit much. Stroke order is almost as important as what the actual strokes are in the definition of a Chinese character, for example. Of course unless you literally watch someone write you observe the characters after they're written, but the most predictive latent mental representation of a character does include an order component. I know this because I made the mistake of memorizing many characters almost like bitmaps and have had to go back and learn how to reliably write/read hand written characters.
Discussion groups in small private spaces are a somewhat related but clearly and justifiably different issue from huge kinda-sorta-maybe-public fora like YouTube. Trying to conflate the two will probably just lead to confusion and muddle the discussion.
To continue playing the analogy game, let's say you run a bible study group in your home. Somebody comes in and advocates for free (homosexual) love - do you have the right to ask this person to leave? Obviously. But what does that say about YouTube's rights? Or more importantly, half leaving aside the question of rights, about what they really should do in a robust free society?
Yudkowsky flirts with tautology in the title - "well-kept garden" has keep right there explicitly and "garden" implies a caretaker. So almost by definition a well-kept garden won't persist without intervention.
But also of course not all ecosystems are gardens; not all ecosystems rely on keepers who are external to the ecosystem. A normal forest is sustained by natural forces and the normal actions of its inhabitants. As much as possible I'd like a community that sustains itself rather than one that is gardened.
Speaking of established case law, we have from the very first page of Heller [1]:
> The Second Amendment protects an individual right to possess a firearm unconnected with service in a militia... [2]
The Constitution encodes a value judgement about the rights it protects; these rights are presumed to be protected unless it can be shown why they aren't in some specific case. That is to say, to ask someone "how [their] right contribute[s] to the security of the state" is wrongheaded - if you wish to deprive them of their right you must first meet strict scrutiny.
I wouldn't be surprised if priests or monks had quite long work hours; imagine the RSI one might get from copying out whole books by hand.
You could reframe as underemployment a lot of the extent to which jobs in the past were less stressful (if they actually were). Trading stress for fantastic material wealth is a highly popular option afforded by technology.
This is sort of true, but the graph of citations is basically a DAG - very unlike the graph of hyperlinks on the web. From what I've seen it's not obvious that PageRank on a DAG tells you anything super interesting.
You've shown that many midwesterners are anti-progressive, but that is not the same thing as anti-intellectual. Furthermore, the optics of where they stand relative to young professionals in the sciences are just that - optics that don't necessarily reflect anything. Consider student movements in the past that haven't gone anywhere or were really misguided in hindsight (there are enough).
Semantic Scholar is completely non-monetized (and will stay that way as far as I can tell) because Paul Allen pays the bills and has a public interest goal.
You can't just try and smudge the line between speech integral to criminal conduct and defamation and other speech. The way we avoid sliding down the slope is by requiring an immediate demonstrable harm - any fuzzy argument of the "can cause" variety is too easily applied.
I am a little confused by your formulation of relevance, or at least the "to what" implicit in it. If we are speaking of relevance to predicting the future or steering societal opinion I would say the more reductive an argument the better. Not that that's a good thing, but I don't see many recent counterexamples.
Really the above thought is related to the "worth" of views or arguments. I'm not sure there's more than a tenuous link between many practical worths and specialized study in a college.
Although actually I feel validated in one respect - it sounds like it took a huge amount of data cleaning and human validation to use the click log data for training.At the time we got many questions about why we didn't "just use it". Although now that I'm at a company with several orders of magnitude more traffic I am a believer in the almighty click log.