lemmsjid

Interesting. Things I think of as quite reasonable, though certainly with counter arguments, are to him fundamentally preposterous and not even worthy of reasonable consideration.

My kid goes to a very liberal California school. The main difference between it and my school growing up is that it is no longer acceptable to ostracize or beat up lgbtq kids or kids who are different in other ways. Part of that is because, gasp, the school builds acceptance into the curriculum. I wish I grew up now, it’s such a nicer time to be a weird nerd.

You’re ignoring general heavy workloads such as observability. How much telemetry do they gather and analyze for tracking and fraud detection. A quick google on Tesco engineering shows that they process 35k qps against couchbase, and 35 terrabytes of telemetry data per day. They track 150k devices in their ecosystem, which, reading between the lines, would produce the telemetry and require observability, state management, anomaly detection, etc. They have hundreds of thousands of employees using these devices for varying purposes. We’re talking quite a bit of compute, which also requires high availability.

I know nothing about Tesco beyond that quick google search, but I’ve been at several companies where I would read online comments claiming we could reduce our workload to a few servers, and I would think of our tens of thousands of fully loaded machines and roll my eyes.

I’d say a useful way of thinking about caching is through the lens of the CAP theorem. You are facing a situation where compute requirements exceed the bounds of a single process. There are a variety of things you can do here, all with consequences to the Consistency aspect of your data. Two strategies with consequences are caching and horizontal scaling. So look to vertical scaling or efficiencies in data modeling first.

I like your comment btw. I’d add Observability to CAP to incorporate what you’re saying.

Quite agree, this is how I explain it to people. When you think of cache as another derived dataset then you start to realize that the issues caches bring to architectures are often the result of not having an agreement between the business and engineering on acceptable data consistency tolerances. For example, outside the world of caching, if you email users a report, and the data is embedded in the email, then you are accepting that the user will see a snapshot of data at a particular time. In many cases this is fine, even preferred. Sometimes not, and instead you link the user to a realtime dashboard instead.

Pretty much every view the user sees of data should include an understanding as to how consistent that data is with the source of truth. Issues with caching (besides basic bugs) often come up when a performance issue comes up and people slap in a cache without renegotiating how the end user would expect the data to look relative to its upstream state.

I enjoy wordle and wouldn’t automate it to replace my play. I also enjoy techie people demonstrating that you can automate X with Y tool. That’s another form of problem solving. Can’t those enjoyments exist simultaneously?

Huh, I agree with your last sentence but think the author did a good job of explaining that the cost of layout experimentation in a startup can grow over time if the results of the experiments are overstated. The first question for a startup should always be: is this work worth doing in the first place? Tinkering with layout can be a tempting but fruitless rabbit hole. Even if it doesn’t tie up resources it can lead to a false sense of progress and get product thinking stuck in local maximae.

As an ex new englander I approve of y’all over youse because it’s hard to use “youse” efficiently without saying “youse guys”.

I partially agree with you but have some counter thoughts.

Tone is something that can be adopted intentionally or unintentionally. If you hear a pilot on a radio dryly say something in a calm and detached tone, it could be in the context of an emergency. Pilots are enculturated to adopt that tone (for various reasons). Meanwhile particular cultures have different levels of acceptability when it comes to tone: some cultures perceive other cultures as more angry, or detached, because of the norms of communication within those cultures.

In short, I think the tone of “calm, scientific detachment” is often weaponized to lend undeserved credibility to an argument, because people tend to believe people more when they adopt that tone.

Furthermore, tone does have a purpose if used alongside a well done argument. For example, in the article the OP linked to, there is a rather exhaustive refutation of the book in question. The tone of the author previews that their entire opinion on the book is negative, given all the arguments they put forth in their review. If the author of the review had adopted a calm and thoughtful tone, perhaps it would indeed have been more effective because the reader would decide. On the other hand, most people won’t read the entire review, so the tone of the author makes it clear what their opinion is.

That said I am not wholly disagreeing with you: would be interesting to do a study using some varying markers to identify tone, and identify, I don’t know, argumentative complexity, and see if snarkiness is associated with a lack of complexity. Assuming you can find markers with predictive power.

I was intrigued and took a Quick Look at the top studies on this subject and the metrics used are things like relative overdose deaths in an area, crime statistics, and usage of treatment programs. They say that by virtue of a number of epidemiological metrics that safe consumption sites appear to be associated with harm reduction in terms of overdoses, while not increasing crime stats. I don’t see outsized claims of objective truth being made, more of the standard, “here’s how we got the numbers, here’s the numbers, they appear to point in this direction.”

I’m not doubting your claim but I’m wondering how that very weird paper you’re citing bubbles up to the top, when there’s some very middle of the road meta analyses that don’t make outsized claims like access to objective truth.

Personally, I find a team seems to be healthiest if a mixture of personalities are on it. They should all be competent.

When it comes to incompetence, people can hide that behind kindness, or behind bluster and bullying. Both are certainly unhealthy. I wouldn’t say one leads to issues over the other.

Huh! If I reflect back on my involvement in projects that had difficulties, there was rarely a dearth of competent people, and in fact it was often political and communication concerns that led to suffering.

Look at Conway’s Law: “any organization that designs a system (defined broadly) will produce a design whose structure is a copy of the organization's communication structure.”

The “kind” people are the people who optimize an organization’s communication structure by helping competent people to have a voice and not be impeded by political wrangling.

In short, I think it’s the ‘kind’ people who can help an organization realize an architecture that is less warped by political considerations and more true to the customer’s needs.

Of course an organization needs both kindness and competence. In my decades in tech, competence was over-valued in my early years (the worship of the trope of the rockstar-but-asshole programmer), so if there is an overindication towards kindness right now, it is probably a counterbalance.

I would also question your conclusion about the government. While I have not worked in the government myself, I come from a sort of “federal government family”, in that I have multiple close family who have spent decades in federal government roles, and they are full of stories of incompetent managers undermining their employees, politically fighting one another, etc. To your point, they also have plenty of stories of crass incompetence, Nepotism, etc. But I think it’s an easy and incorrect answer to say it’s simply due to “HR-values” as opposed to “engineering-values”: it’s multi-faceted in both directions.

I think what you're seeing there is Jemisin actively foregrounding and exploring a difficult fact, which is that people who were abused are more likely to turn around and be abusers.

Often times books turn abuse into the catalyst for a noble struggle and cathartic improvement, when in fact that often isn't the case. There's large swathes of human history where oppressive regimes were powerful enough to quickly and brutually suppress rebellions. When violence and abuse are normalized in a culture, it can take a long time and a lot of failed struggles to truly unwind the violence.

I think Jemisin is actively trying to explore that mentality, which is pretty brave because it does make the characters off putting. As the reader you want so badly for them to rise out of the muck and take on a noble struggle, but they're caught up in the cycle of violence.

I say this with respect for your viewpoint, I think you're being very fair about expressing your response, and I completely see where you're coming from.

On your first point, you have to at least acknowledge that a ton of people strongly disagree with you. I've certainly bounced off of widely read and highly rated novels before, but rarely do I leave with a complete dismissal of the quality of the work. It seems almost personal to you, calling a widely and multiply published author a "child throwing out random ideas".

On your second point: The Nebula Awards specifically target "science fiction" and "fantasy" genres. I'm quite widely read in science fiction. I took a look at the list of novels that have won or been finalists for the Nebula Awards. It turns out I've read quite a few of them, scattered over the decades the Nebulas have existed. The Fifth Season is far from the only set of books with fantastical elements to have won, and the ones I've read that have won have plenty of world building elements that would be incorrect by a fourth grade textbook (FTL? Reincarnation? Fire breathing dragons?). Yes, the Game of Thrones novels (a Song of Ice and Fire), though quite excellent, are quite firmly in the realm of fantasy, and they were often Nebula finalists! By your criteria, the Nebula Awards were "destroyed" a long time ago. Several books, including some from the very early years of the Nebula Awards, are now called "science fantasy", i.e. science fiction that leans into fantastical elements. The Fifth Season is actually typically labeled as "science fantasy".

If the Fifth Season "destroyed" the Nebula Awards, then you would see more recent winners being almost exclusively science fantasy. I haven't read as many of the newer winners as I have older winners, but two of my favorite novels of recent years won or were finalists: Network Effect by Martha Wells, which is pretty much a core science fiction novel, and Piranesi, by Susanna Clarke, which is even more fantastical than the Broken Earth.

One of my favorite recent-years science fiction novels, Children of Time, surprisingly didn't get a Nebula, but it did get a Hugo. Looking at its publishing year though, 2015, I see why! Three Body Problem was nominated, and Annihilation, which I absolutely adore (but can't spell), won that year.

No flamewar here, I quite agree! I really enjoyed The Fifth Season (if "enjoy" is a word that can be applied to something so dark), but it did have flaws (I used "visceral" and "imaginative" in my above comment because I think those are her strengths: I really felt immersed in her descriptive writing).

Meanwhile Octavia Butler is absolutely one of my favorite authors regardless of genre. Her work stuck with me for quite a while, as has le Guin's (specifically, Left Hand of Darkness, I've actually had trouble getting into some of her other works, though I mean to try again). Butler to me is a visionary alongside my other favorites in the sci fi canon.

Interesting. I've made it a point to try to make my way through the canonical sci fi greats over the years, and I'd put Left Hand of Darkness near the top. It's kind of because of that tonal difference: there's something about the way it's written (sort of anthropological / travelogue style) that makes me feel truly immersed in the culture and world being described, in a trance-like way, even though the book itself doesn't have particularly exciting events in it.

It may be that it hit me at a particular time in my life: I read it in my early teens and the way gender was expressed in the novel, the sort of tidal shift between masculine and feminine based on circumstance, really spoke to me at a time I was figuring all that out in my own psyche. I wonder if a lot of the gender exploration in the book may seem more trite and typical now.

Jemisin is one of the more visceral and imaginative writers I've encountered in recent years, and she did indeed produce and publish written works, so I do believe she is a real author, yes. I'd certainly be interested in reading and maybe discussing an actual critique, this being a forum where substantive posts are required in the guidelines.

I wouldn't say it's a crude way of approaching the problem, it's a crude way of solving the problem. Taking the fraud example, taking unsupervised approaches to understanding patterns of the data before you impose assumptions on the data is a very useful process. For example, what might be fraudulent behaviors in the first place, assuming you aren't even sure you know what fraud looks like, or that it's actually all been detected? Your goal there might be to detect latent features period, not look at their predictive power for X.

Having understood that question, and built an understanding of what predicts fraud, you would then graduate to build models to understand the extent to which features predict fraudulence.

My point in context of the conversation is that it's useful in a business context to explore and understand that data.

I might be misunderstanding your point, but there's use cases that have repeatedly come up for me in multiple businesses, below being some examples, without getting too specific:

- identify latent features of customers via their behavioral data, to be used for profiling customers or recommending products to them

- within a large amount of customer behavioral data, identify potentially fraudulent behavior

- identify causes of seasonality (e.g. temporal patterns) in the data in order to improve forecasting (sales, traffic, whatever)

In those cases part of the investigation is to initially take a hands-off (unsupervised) approach, so that we can compare our initial top-down hypotheses with actual patterns in the data.

In both of those cases there's considerable (and sometimes adversarial) noise in the data.

Poor choice of words on my part. In my example one is going "off course" in the sense that one is spending more time on the yak shaving than is justified by the scope of the problem being solved (e.g. let's say refactoring a script that is only run once, and spending more time on the refactoring than if one had manually replicated the script's output). Maybe "too deep in the rabbit hole" would be a more illustrative phrase.

I've heard the term used in a purely negative sense, while personally thinking it's often necessary and sometimes brilliant to do some yak shaving. Some of the best yak shavers I know are essential people, even in a "move fast" startup environment, because they effect significant change and improvement that can lift the whole team out of a local maximum of productivity.

Maybe for people who use it negatively, "refactoring" or "unplanned improving" isn't "yak shaving" until you've gone beyond what's reasonable. E.g. if you're doing some recursive refactoring, and improving the code base, great, but yak shaving is when you've truly gone off course.

Either way, I think this is where healthy team dynamics are so useful, because it's hard for the yak shaver to know if they've gone off course or are doing something unplanned but useful. If I think I'm deep in a yak shaving exercise, but I think the results will be beneficial, I announce that I'm yak shaving and explain why. If the rest of the team thinks it isn't truly a useful exercise, they can guide me out of it.

It would be interesting to do an experiment on people playing one another (online) with these pieces vs regular pieces (e.g. one player with a regular set and one player with this set). Because I do feel like I can visualize the state of the board more easily with this approach--but would that translate across people into better play, or is there some point where a more expert player (I don't play very often, though I enjoy playing and have played for years) has already internalized the mapping from piece shape to movement. But that said, it does feel like the difference between font styles in programming, which for me have a very meaningful impact over time.

Edit: Though good point to the parallel commenters, the knight shape is harder to differentiate and kind of throws me off. But maybe tweaks there.

I just finished reading Children of Time and was indeed imagining that he was looking at earlier research on this subject, because the way he describes the spiders' initial thinking skills quite echos the article. It was very fun to read from that perspective!

It is not racist to make an observation about race. It is racist to believe that one race is superior to another and act accordingly. For example, it is not racist to acknowledge the frequently measured metric that African Americans on average score lower than other races on scholastic aptitude tests. It is what one does with that observation that can lead to racism.

It is useful to define terms. Merriam-Webster has what I believe is a quite useful definition of racism: "a belief that race is a fundamental determinant of human traits and capacities and that racial differences produce an inherent superiority of a particular race".

That is a useful definition because it helps to separate a racist interpretation of test scores from a non-racist interpretation. The racist interpretation is to jump to a conclusion of fundamental inferiority in academics. The non-racist interpretation is to acknowledge and emphasize the environmental factors that can lead an historically oppressed/displaced/enslaved minority to underperform academically, to acknowledge that African-born immigrants do quite well on standardized testing, and also to understand that standardized tests are normed on academic success, which creates a causality issue when measuring groups that have historically not been allowed in academic institutions. Perhaps all of those arguments don't hold water: or at least the pitcher holding them has holes of its own. But it is quite erroneous to jump to a conclusion of racial biological determinism, and that is what the racist conclusion would be.

There is a particular kind of fallacy where a person can, seemingly unconsciously, take what are perfectly valid observations, like the test-taking disparities, and spin them into non-evidential overarching beliefs. This happens all too commonly when it comes to scientific (or at least metrical) observations and racism. People were all too eager to embrace phrenology, IQ testing, evolutionary biology, etc. when it confirmed their cultural preconception that their race was superior. All of those areas, even phrenology, had some scientific merit to them (and still do in the case of IQ testing and evolutionary biology), but that merit is lost when the conclusions get spun into non-evidential belief systems like racist ideology. Seemingly (I can't peer into their heads, really) skepticism is lost in a chase for the personal gratification of a belief of fundamental superiority.

Discussions around academic standards are a case in point. Academic standards change quite frequently. The Oregon requirement we're discussing was only instituted in the early 2010's. When race is not part of the discussion, it is entirely normal for educators to say that a new standardized test doesn't really have evidence it serves the population and to remove it. There is a very lively debate amongst educators about how excessive measurement, i.e. standardized test taking, can interfere with the normal give-and-take of teaching. But, if the evidence is that non-native-English speakers are underserved, or that African-Americans are underserved, suddenly it is 'racist' to remove a requirement or hurdle. What I saw in the two articles is that those in favor of removing the requirement saw a lack of evidence it was beneficial, and saw evidence that it was not beneficial; while those opposed had no evidence other than it would graduate 'dumber' students; ignoring the fact that presumably dumber students were being graduated before the 2010's when the program was instituted.

In short, it is often a smokescreen to say that observations about race are racist, because that makes it impossible to act against inequality, or sometimes even to act at all. Once again, racism is the belief of one race's inherent superiority over another.

I am as layperson as they get on this subject matter, but I would think that even if your statement is true, perhaps we can get to the point where bacteria need to evolve so much to overcome new antibiotic approaches that they lose some of the properties that make them harmful and transmissible, i.e. it becomes harder for them to exist outside of the host, penetrate the host's defenses, etc. At some point a bacteria would need to seem so like a human cell or beneficial bacteria that it becomes non-harmful.

Ah good show :). I was rather preoccupied with the question but didn't have one handy. Well, I do, but my kid would roast me slowly over coals if I so much as smudged it. Ah the joy of the Internet, I did not predict this morning that I would end the day preoccupied with the question of rubber duck density!

I guess for me the question of whether or not the model is lying or hallucinating is if it's correctly summarizing its source material. I find very conflicting materials on the density of rubber, and most of the sources that Google surfaces claim a lower density than water. So it makes sense to me that the model would make the inference.

I'm splitting hairs though, I largely agree with your comment above and above that.

To illustrate my agreement: I like testing AIs with this kind of thing... a few months ago I asked GPT for advice as to how to restart my gas powered water heater. It told me the first step was to make sure the gas was off, then to light the pilot light. I then asked it how the pilot light was supposed to stay lit with the gas off and it backpedaled. My imagining here is that because so many instructional materials about gas powered devices emphasize to start by turning off the gas, that weighted it as the first instruction.

Interesting, the above shows progress though. I realized I asked GPT 3.5 back then, I just re-asked 3.5 and then asked 4 for the first time. 3.5 was still wrong. 4 told me to initially turn off the gas to disappate it, then to ensure gas was flowing to the pilot before sparking it.

But that said I am quite familiar with the AI being confidently wrong, so your point is taken, I only really responded because I was wondering if I was misunderstanding something quite fundamental about the question of density.

I did some reading and it seems that rubber's relative density to water has to do with its manufacturing process. I see a couple of different quotes on the specific gravity of so-called 'natural rubber', and most claim it's lower than water.

Am I missing something?

I asked both Bard (Gemini at this point I think?) and GPT-4 why ducks float, and they both seemed accurate: they talked about the density of the material plus the increased buoyancy from air pockets and went into depth on the principles behind buoyancy. When pressed they went into the fact that "rubber"'s density varies by the process and what it was adulterated with, and if it was foamed.

I think this was a matter of the video being a brief summary rather than a falsehood. But please do point out if I'm wrong on the rubber bit, I'm genuinely interested.

I agree that hallucinations are the biggest problems with LLMs, I'm just seeing them get less commonplace and clumsy. Though, to your point, that can make them harder to detect!

I think atomic bombs is a good analogy for why it's alarming, actually. Humanity had a great and brief time putting radium in all sorts of things until they realized that it was doing something that penetrated deeply into and through the body at a micro scale.

Humanity has also had a great time making different kinds of plastics, only to discover that they are also penetrating our body at a microscopic scale.

While we don't have the smoking gun yet (i.e. radium jaw), it's at least scary to me, because platics are so pervasive across the population that if a smoking gun DOES emerge, it might be too late. I think the article itself brings up another good example: asbestos.

I get all the jokes about dihydrogen monoxide killing people who breathe it, i.e. drowning, but this is fundamentally different, because we are discovering a thing that is interacting with our bodies in a way that was not historically present throughout evolution. Presumably if we'd evolved in the mist of nano plastics there wouldn't be much cause for worry, because a priori we'd have survived and thrived in its presence.

I think you're really re-stating his case, which is interesting because you then call it 'empty virtue signaling'.

He's applying a standard analytical lens to AI startups, e.g. looking for their moats through finding differentiators in economics, data, scalability, etc. He finds that "doing AI" is not a stable enough differentiator to compel him as a VC. He then lays out his reasons. There are plenty of startups selling themselves on their AI platform and/or acumen, so it's rather automatically relevant to a VC at least.

I'm a supporter of free speech, and like most free speech supporters I tend to believe the suppression of speech leads to a paradoxical effect: the metastasis of the suppressed ideas.

Of course there is a but coming! When twitter launched, I immediately felt like Twitter itself was a distillation of a fascistic impulse: the idea that a curt and pithy phrase, stripped of nuance, is not only a superior way to communicate, but the only way to communicate. (I'm using "fascistic" here not to incite, but for lack of a better term: I think everyone has an impulse towards the safety of political simplicity, and that modern democracy is founded on complexity, i.e. lots of opinions, lots of checks and balances, lots of persuasive argument required to achieve things). I feel that the very structure of throttled meaning is what made Twitter in particular such a platform for mass outrage and negativity.

Fast foward a decade, and the twitter model of simplicity is now the normal mode of discourse on social media. If something is not a few seconds long, its reach is quite limited. This means that simple things, like propaganda, are easier to articulate, and complex things that acknowledge the subtlety of actual life get buried.

Taken out of political context, a good example is videos of altercations between people (fights, arguments, etc) that become viral. As a viewer, I almost immediately feel outrage and righteous anger against the perceived victim of the altercation. But sometimes some context comes to light: the perceived victim is actually the aggressor, or the situation is much more complicated than it seemed. Putting this back in the political context, I see quick videos getting passed around that do a similar thing: take an event quite out of context in order to manufacture outrage against a particular group. Of course there is a word for this: propaganda.

I'm not a technological doomer: I'm quite intrigued as to what is happening, and my impulse is not to shackle it. But what I'm witnessing is that propaganda (or otherwise, events taken out of context) spread like lightning on social media, and I am sympathetic to attempts by platforms to somehow provide context or moderation to that spread. The mob mentality is real. I share the Mayor's concern about Twitter, because even though it was a sewer from the start, in my estimation, by its platformization of epistemological constraint, the fact that it is walking back attempts to staunch the spread of its own demons is concerning.

I do think a healthy public debate in a functioning, free speech democracy, is essential to its lifeblood. But for that debate to function, it must be done in good faith, and with some shared understanding of truth and context. While I see that happening in some places on social media, it's generally on well moderated forums in which people accept that bad faith arguers will be banned.

Like I said I don't have a clear sense of the solution to all this. State sponsored propaganda was all too effective before social media, and I can see social media undermining the monopoly of the state on the dissemination of that propaganda. But perhaps the pendulum will swing too far to the other side, where blatant bad faith, grassroots propaganda, and untruth successfully steer public opinion.

To me this blog post doesn't fully make its case, though has many good points and is a good read.

I think my main logical objection is that the alternative best practices at the end of the article were all security best practices before WAFs existed. Which makes me ask the question, why did WAFs come into existence in the first place? Did the founders of those companies convince customers they needed them without those customers actually need them?

I think not. In the years before WAFs existed, I was in the position more than once of being in an organization whose web application security footprint had grown to the point where we ended up writing a home-grown version of a WAF. E.g. adding an interception layer that would analyze inputs and outputs for typical security violations.

Why? Well, first because it started to give us a sense of the types of attacks that people were trying to use. Second, because the types of mitigations mentioned by the author of this blog post aren't the whole story. You can audit that your entire system avoids SQL injection attacks via stored procedures, then your company buys another company with a code base that fails such audits. Or someone attacks by leveraging your caching layer which stores and sends back unaudited key-value pairs. Perhaps (this has happened to me) a bug gets introduced into the deployment system, and the code that forces authentication is not shipped, and the calling code doesn't properly fail when the auth checking code isn't in there. A real head slapper in hindsight.

I do like the best practice of process isolation around APIs, and only allowing APIs to have the privileges they require, but in practice, if there are hundreds of APIs undergoing frequent changes, the complexity of managing that becomes a security risk in and of itself, because the ACL rules are deeply complicated.

Relying solely on a WAF seems like a bad practice. But also relying only on secure design philosophy is a practice with plenty of historical failures.

So if the point of the article is that WAFs breed complacency, I agree with that! But if a WAF is used as an analysis, auditing, and fast-response layer, alongside following secure design principles, then I'd say that based on personal experience, if WAFs didn't exist, people would write home grown ones with their own sets of flaws.