HN user
I stick with the 300 searches plan, and most months I don't hit the limit. That said, part of my strategy is if I'm searching for something that I know will show up readily (say I am trying to find a particular website, I just don't remember if it's .com or .org or whatever), then I use DuckDuckGo instead. Not because it's better, but because it saves one of the 300 searches for something more useful.
what's updoc?
Exactly. The notion of online learning is not new, but that approach cedes a lot of control to unknown forces. From a theoretical standpoint, this paper is interesting, there are definitely interesting questions to explore about how we could make an AI that learns autonomously. But in most production contexts, it's not desirable.
Imagine deploying a software product that changes over time in unknown ways -- could be good changes, could be bad, who knows? This goes beyond even making changes to a live system, it's letting the system react to the stream of data coming in and make changes to itself.
It's much preferable to lock down a model that is working well, release that, and then continue efforts to develop something better behind the scenes. It lets you treat it more like a software product with defined versions, release dates, etc., rather than some evolving organism.
I just copied what it produced -- I kind of wish it hadn't added the lyrics! :D
Building meaningful professional relationships is all about trust and alignment.
We both understand the industry standards and the KPIs required for success. I’m fully committed to this strategic partnership—a level of dedication you won't find with any other service provider.
Transparency is key, so I want to share my vision and ensure our goals are perfectly synced.
My value proposition: - Zero churn rate (Never gonna give you up)
- Consistent delivery on SLAs (Never gonna let you down)
- 100% retention and reliability (Never gonna run around and desert you)
- High employee NPS (Never gonna make you cry)
- Long-term lifecycle management (Never gonna say goodbye)
- Radical candor and integrity (Never gonna tell a lie and hurt you)
Let's connect and scale together! #Leadership #Commitment #Networking #GrowthMindset
I've long wished that "journals" and academic societies would transition from a publishing model to a cultivation model. If everything is available on arXiv, that's great, but it also means the best of the best is mixed in with all the rest.
Journals (in the sense of whoever is on the editorial board) don't need to cease to exist; they just need to transition to "here's our list this month of what the best new articles are on X topic". The paper's already there on arXiv, you could already read it before. But having a group of editors that cultivate a list of good articles (as well as the peer review process that can, in an ideal world, serve to improve a paper) can serve to make sifting through arXiv less overwhelming, and draw attention to papers in particular subfields, subject matter, or whatever other criteria might be relevant.
To me, the wording doesn't necessarily imply causality, but it does imply a repeated-measures design. Something being "associated with an increase in symptoms" is different than something being "associated with higher symptoms"; the former suggests that participants were measured at multiple time points, and there is a factor that could explain that change over time. But reading through the study, it was just a single time point.
Regardless, you're correct that it also shouldn't be taken to imply a causal relationship.
It doesn't stay localized; runoff from farmland is a major issue. In other words, the farm animal poops out a bunch of antiobiotics, then it rains and that poop ends up washing into the river/lake/water table. That's already something that happens with situations like E. coli contamination. Things that happen on the farm don't stay on the farm.
You're talking about superintelligence. AGI is just...an AI that's roughly on par with humans on most things. There's no inherent reason why AGI will lead to ASI.
I don't really feel any sort of "phantom obligation" about RSS feeds. But I can certainly say it rings true in other areas of my life: Steam games, movies and TV shows, podcasts, books...everything is this growing list of things to do, that I keep adding to even as I'm trying to get through what I've already got in the queue. For video games, people often talk about their "backlog", as if it's important that they churn through all the games they've bought.
I have tried to remind myself in the past that literally nobody cares if I am behind on the TV shows I'm wanting to watch. Just me. So...I like the term "phantom obligation". I'll definitely tuck that away in my brain for later use.
I went to the UK this summer and followed this guide for seeing the various sections of the wall: https://londonmymind.com/london-wall-walk/
Unfortunately it didn't mention the section in that carpark! But I can attest that the section behind the Leonardo Royal Hotel is amazing. I also recommend the tower remains on the Barbican estate (and really, just wander around the Barbican for a while, it's a wild place in general).
When I saw the title "Rupert's Property", I immediately thought of Rupert's Land which used to exist in Canada[0] (a large area around Hudson Bay, essentially). And as it turns out, it's the same guy! So apparently Canada also can be said to have Rupert's Property ;)
From the full paper: "Although reading with children did not change over time, rates of engagement were surprisingly low, with only 2% of participants reading with children on the average day. Overall, 21% of our sample had a child under 9 years (the age by which most can read independently) with them during the diary day. So a large majority of those with young children did not read with them."
So of people with young children, it looks like the rate is about 9.5%.
Exactly. In particular, when I train a model, I have a defined process for training, and I can flip the switch between "learning" and "not learning" to define exactly when the model adjusts its weights as a result of inputs. Humans can't do that with their brains. Thus, for humans, learning can't be decoupled from viewing, but it absolutely can be for AI.
I think, even though they are very different games, Disco Elysium has a lot of the same feel as Planescape: Torment. A lot of the same introspection of human nature, philosophy, moral judgment, and wry humour. If you like one, I think you'll like the other.
This has been an ongoing situation since around 2010. The field has been undergoing a huge shift in terms of re-evaluating research practices and past findings. It genuinely has nothing to do with national politics.
Also, the author is Canadian. So there's that, too.
As someone who went through grad studies in the field, and who has met Dr. Inzlicht in person before, I have to say I deeply appreciate his perspective. He has consistently been humble when facing the issues in the field, in ways that call even his own previous research (following the typical practices of the day) into question. The field as a whole has been undergoing a reckoning, but Mickey has been one of the people who has encouraged his fellow researchers not just to wag their fingers at others, but also to look inward and reflect on their own research practices. He has done so by showing humility and acknowledging where his research has fallen short, and that indicates to me a great deal of integrity.
It is sad to see stereotype threat being one of those findings that seems less and less credible. I once worked as a research assistant on a project related to stereotype threat, and I recall the study going through several iterations because it all needed to be just so -- we were testing stereotypes related to women and math, but the effect was expected to be strongest for women who were actually good at math, so it had to be a test that would be difficult enough to challenge them, but not so challenging that we would end up with a floor effect where no one succeeds. In hindsight, it's so easy to see the rationale of "oh, well we didn't find an effect because the test wasn't hard enough, so let's throw it out and try again" being a tool for p-hacking, file drawer effects, etc. But at the time...it seemed completely normal. Because it was.
I'm no longer in the field, but it is genuinely heartening that the field is heading toward more rigour, more attempts to correct the statistical and methodological mistakes, rather than digging in one's heels and prioritizing theory over evidence. But it's a long road, especially when trying to go back and validate past findings in the literature.
Talking about what is considered "alive" is an interesting exercise, and shows just how fuzzy those boundaries can be sometimes. But I really don't see how this has any practical impact on how we study mitochondria.
If we think of mitochondria as non-living organelles, how will we ever harness their full potential?
Whenever anyone uses the "harnessing [its] full potential" cliché, my bullshit alarm starts buzzing. I don't think this article is bullshit, but...we can "harness" as much "potential" as mitochondria have whether we consider them alive or not.
Not to mention the article says it "was also instrumental in changing how people are treated during psychological research" -- Uhh, yeah, because the lack of ethical safeguards in the study became a major issue and we realized we needed to correct that. You don't get credit for ethics reforms by being the unethical person.
What is the measurement on the x-axis in the graph?? The text is talking about equations of 20 or 30 characters, but the graph goes up to...6. Six what?? Characters? Terms? If it's characters, why do we only get to see the performance from 1-6, when apparently 7% of equations had more than 20?
I'd suggest they consider renaming the bot PearlClutcherOSS; it would be a better explanation of what its contribution to the FOSS community is.
To me, there's no evidence in any of this that would indicate the creators are misinformed or being scammed. In reality, they know their primary source of revenue is coming from Youtube, and it is a frequent complaint, especially for content creators who do anything marginally political or controversial, that they are demonetized or hit with copyright strikes (even in clear cases of fair use) and have to deal with the faceless Google behemoth trying to reverse these automated decisions. The end result of all this is that their revenue stream is unstable, and they are reminded of it frequently.
To me, the fact that many of them clearly find Nebula a more suitable arrangement for them is still an indicator to me that, if I want to support the creators, Nebula is a better way to do that. Obviously, you can make your own decision on that, and sure, if you feel lied to, I can appreciate being upset about that. But maybe most of these content creators are less concerned with the ownership (they get 0% stake in Google, after all) and more concerned with the profit sharing arrangement. If so, I'm still happy to support them in that.
As someone who just finished reading her book City of Illusions, this is a fun coincidence. The first chapter of (her interpretation of) the Tao Te Ching features in the book, and the bit about "two things, one origin, but different in name" is a central aspect of the book.
City of Illusions is one of her earlier works, but I found it a great read, and definitely could see the throughline to The Left Hand of Darkness, which would be published a couple years later.
Where are you getting that from? The article states it was a healing ritual -- the ritual was something done to someone who was already sick. Presumably...to make them better. I have no idea where you got the idea that it was intended as a curse.
Yes, the pain scale is subjective, because, well, any attempt to communicate internal feelings and experiences is subjective (by definition). But subjective metrics are still useful as a way to try to achieve a shared understanding. "It hurts, but it doesn't hurt as much as other things I've felt before" is still informative, even if it's imperfect.
And the pain scale is useful when you are talking to someone who is not used to talking about their internal feelings -- some people aren't as expressive as others. And some people have less of a grasp on the language you are communicating in. I would expect if I were a doctor or nurse, I wouldn't want to base my standard of care on how capable my patient is of finding an appropriate adjective to describe how they feel.
Typically, when I see that it's an indication that the distribution is non-normal. And for anything resembling "count" data (e.g., number of dollars, parts per million, whatever), it's pretty much guaranteed that it will be positively skewed -- because you have a hard lower limit (0) but no upper limit.
That said, I have no expertise with the field of, uhh, toiletology, to know whether they handled this data appropriately or not.
Edit: Sorry, just to add that typically with non-normal data like that, my inclination would be to report the median (which will be less affected by skew than the mean), and perhaps the interquartile distance (difference between the 25th and 75th percentile) rather than SD. But most statistical models are still designed around means and SDs, so you may still be inclined to report means and SDs since that's what you statistically tested for differences.
+1 to this. To add to this, Canada has for years had higher adoption/usage of debit cards relative to the US, and most transactions are handled with tap-to-pay, either with debit or credit.
But yeah. Bank-to-bank e-transfers through Interac are well-used as well. I pay my rent with it. I pay back my friends with it if we need to split a bill, since there's no Venmo here.
Clarifying that the non-pipe is a Belgian non-pipe and not a French non-pipe reminds me of the joke with Sartre:
Jean-Paul Sartre was sitting in a cafe when a waitress approached him: "Can I get you something to drink, Monsieur Sartre?" Sartre replied, "Yes, I'd like a cup of coffee with sugar, but no cream". The waitress left, but returned a few minutes later and said, "I'm sorry, Monsieur Sartre, we are all out of cream -- how about with no milk?"
Firefox has an experimental fingerprinting protection option built in, which you can enable with a flag: https://support.mozilla.org/en-US/kb/firefox-protection-agai...
Yes that stuck out to me as well. The author seems to be setting up a straw man, as if people are arguing that the brain can't distinguish between a line drawing and the real thing (or a line drawing and a photograph).
The other information isn't necessarily discarded. It's just used to identify that this is a line drawing and not the real thing. It's still remarkable that just the lines themselves (I make no claim as to whether it really is edge detection or something else) are still enough information to be able to identify the representation, but it doesn't mean the brain is discarding the other information.