f38zf5vdt

As soon as I saw the reported murder I figured it was a business partner who felt scorned, but then HN narrative continued for a full week about the lack of safety in SF. While things might be bad, random murders by strangers are still a rarity compared to targeted killings.

I know it is hard being a forum BDFL and it's expected that we will not always agree. Seeing a lot of unsubstantiated medical advice during the pandemic was just very hard for me, especially when hearing it from a "smart" place like HN. For better or worse HN tows the line on US-style free speech of letting bad but eloquent actors promulgate their ideas for fear of censoring the next Galileo.

Eat it dang, you wouldn't let me poke fun at morons recommending horse dewormer for COVID19.

I hate living in the future.

No one wants to go to jail for Elon, who has been flagrantly violating FTC orders.[1] There's a good chance the commit history and authors may attest to that.

https://thehill.com/policy/technology/3928219-musk-was-denie...

It's a proprietary versus open source argument. You can build a mk4 clone for much less than the price of a Prusa-branded one, as everything is open source. The higher price tag is like a donation to open source.

Prusa also makes the open source XL if you want an CoreXY model, or you could go with a Voron which is also open source.

> In general any type of hormonal anything is probably dangerous. Endocrine system is incredibly complex and connected in many different, mostly unknown ways.

By that logic sugar should be a controlled substance and we should all be gonadectimized. Gonadectimized animals almost always live longer.

> And contraceptives play with hormones that are central to what makes women women.

Weirdly estrogen in utero is central to what makes male mice male mice, and adult men adult men (20–55 pg/mL estradiol in adult men or about 1/10th the level of adult women). Estradiol plays many important roles in the body of all human adults.

> These hormones are not just limited to regulating when eggs are released from ovaries, they take part in a lot of other processes. It is absolutely not surprising then that there will be other effects of long term hormonal supplementation.

> I think the best course of action would be to work on other, non-hormonal contraceptive alternatives.

Maybe, maybe not, since chronic use of any medication will result in chronic side effects and physical interventions such as copper IUDs are also associated with a host of serious side effects. The best course of action would be to develop interventions that are safe in animal models, give them to humans for a while, and monitor the outcomes, which is what is going on here.

I'm surprised that the outcomes observed aren't worse, given how strongly carcinogenic estrogens are on a dose-dependent basis in animal models. Hormones are strange. Women with many children have reduced amounts of breast cancer despite being exposed to absolutely enormous levels of estrogens and progesterone during and after pregnancy.

...what even is SushiDAO? A Google search for "What is SushiDAO?" just gave a lot of articles with crypto-word-soup.

You are humanizing token prediction. The multimodal models for text-vision were all established using a scaffold of architectures that unified text-token and vision-token similarity e.g. BLIP2. [1] It's possible that a model using unified representations might be able to establish that the set of visual tokens you are searching for corresponds to some set of text tokens, but only if the pretrained weights for the vision encoder are able to extract the features corresponding to the object to which you are describing to the vision model.

And the pretrained vision encoder will have at some point been trained to minimize text-visual token cosine similarity on some training set, so it really depends on what exactly that training set had in it.

[1] https://arxiv.org/pdf/2301.12597.pdf

Image classification is still a difficult task, especially if there are only a few examples. Training a high resolution 1k multi-class imagenet on 1m+ images is a drag involving hundreds or thousands of GPU hours from scratch. You can do low-resolution classifiers more easily, but they're less accurate.

There are tricks to do it faster but they all involve using other vision models that themselves are trained for as long.

Weird. It has no effect on survival in rats and the doses were large.

> One chronic (78-weeks) study in rats with dietary levels of 0, 1, 3, or 10% erythritol (equal to > 0.46, 1.4 and 5 g/kg bw/day for males and 0, 0.54, 1.7 and 5.7 g/kg bw/day for females) (Til > and van Nesserooij, 1994) and another 2-year chronic toxicity/carcinogenicity study in rats > with dietary levels of erythritol of 0, 2, 5, or 10% (equal to 0, 0.9, 22, and 4.6 g/kg bw/day for > males and 0, 1.0, 2.6, and 5.4 g/kg bw/day) (Lina et al., 1994; 1996) demonstrated that > erythritol did not affect survival and had no carcinogenic effect.

https://ec.europa.eu/food/fs/sc/scf/out175_en.pdf

From this paper:

> In a population-based prospective cohort study with repeated dietary records, ingestion of multiple artificial sweeteners (for example, aspartame, acesulfame potassium and sucralose) was associated with CVD risk.

Ok, let's look at that reference.

> Results Total artificial sweetener intake was associated with increased risk of cardiovascular diseases (1502 events, hazard ratio 1.09, 95% confidence interval 1.01 to 1.18, P=0.03); absolute incidence rate in higher consumers (above the sex specific median) and non-consumers was 346 and 314 per 100 000 person years, respectively. Artificial sweeteners were more particularly associated with cerebrovascular disease risk (777 events, 1.18, 1.06 to 1.31, P=0.002; incidence rates 195 and 150 per 100 000 person years in higher and non-consumers, respectively). Aspartame intake was associated with increased risk of cerebrovascular events (1.17, 1.03 to 1.33, P=0.02; incidence rates 186 and 151 per 100 000 person years in higher and non-consumers, respectively), and acesulfame potassium and sucralose were associated with increased coronary heart disease risk (730 events; acesulfame potassium: 1.40, 1.06 to 1.84, P=0.02; incidence rates 167 and 164; sucralose: 1.31, 1.00 to 1.71, P=0.05; incidence rates 271 and 161).

This study shows the same thing with _all_ artificial sweeteners, and has way more people (n=103,388) than the present Nature study (n=1,157). And we know that erythritol is often used by type 2 diabetics due to its low effect on insulin levels. That population is predisposed to cardiac events to begin with.

When adjusting the US/EU cohorts for cardiovascular risk factors, the observed effect with erythritol are smaller and closer to this study of other artificial sweeteners.

> Consistent with the results observed within the discovery cohort (adjusted HR = 2.95 (1.70–5.12) P < 0.001; Fig. 1 and Supplementary Table 4), the association between erythritol levels (fourth quartile versus first quartile) and incident MACE risk remained significant in both US and European validation cohorts following adjustments for cardiovascular risk factors (adjusted HR (95% CI) = 1.80 (1.18–2.77) and 2.21 (1.20–4.07), P = 0.007 and P = 0.010, respectively

I'm not convinced of the validity of this study. They ignored the rat chronic toxicity data and proposed a mechanism for what they saw in a small cohort of humans, but it should be reproducible in mammals.

Spoliation is a crime

https://lewisbrisbois.com/newsroom/legal-alerts/15-month-pri...

Yes, I phrased this badly. I meant that instead of using the phone's NFC for verification, we can use a small NFC device that we tap to the phone and which the phone reads via NFC to verify our identities. Just like we tap our cards to payment terminals.

Yes, a government or corporate issued device like a Yubikey with NFC is my ideal. Just tap to the phone to verify. Payment cards got it right, everyone else is just catching up.

I feel like a broken record, but in the near future we are going to switch to a public key authentication object for everything social we do. Payment has already moved to it: tap payments use public key cryptography.

Yet, at the same time, whenever it is brought up people love to engage in contrarian rambles about how it is simply too hard, people will lose their dongles, etc etc. Most phones support NFC communications these days, it's a tap against a phone or reader for verification and so couldn't be easier.

Within 10 years people are going to wonder how we ever lived _without_ this.

Plausible deniability only gets a country so far. In historical terms state-sponsored terrorism has warranted military action when the severity evolves enough. It's hard to consider British soldiers fighting for Ukrainian sovereignty in the same realm as disrupting children's hospitals. [1][2] I had read that one of the attacks had disrupted imaging systems and prevented diagnoses from being reached, delaying treatments.

[1] https://www.healthcareitnews.com/news/fbi-disrupts-hive-rans... [2] https://therecord.media/canadas-largest-childrens-hospital-s...

If Russia were to physically disrupt a country's social services by force, it would be an act of war. Here they have disrupted a British social service without force by state sponsored hacker terrorists, rendering harm in the physical world. The line gets blurrier.

It's remarkable how close the UK is getting to CN-styled internet. It would not surprise me to see a proposal to firewall the island off from the rest of the world, mandatory British government apps for services, etc., following the permanent installation of a dictator. Coupled with the promotion of a national digital currency, it's hard to see a future where Britain does not devolve into fascism. It has not had a democratically elected leader for years now.

Whatever is readily available to be converted into digital "cash" when needed will be used. Wealthy people maintain most of their assets outside of cash in property of various types. When they need "on the books" spending, they bring it into the cash economy with on/off ramps.

I don't know if the UK is different from much else of the developed world, but here there is a tremendous amount of off-by-book transactions in the largest industries such as farming and construction. I do not think that the disappearance of cash will remove this economy, but it will have to migrate to other assets with similar qualities.

It will cause people to stop using government money. There are many other ways to barter and exchange goods. I expect it will only increase tax evasion as less transactions are even reported.

It's no surprise to me to see government gold buying on an absolute tear. [1] The powers that be are well aware of the importance of having real physical goods for the sake of trading and maintaining wealth.

[1] https://financialpost.com/commodities/mining/central-banks-b...

This is a company that allows you to search for images from a training dataset that have a high cosine similarity with a given image. It appears to be the same as the open source software published by LAION.

https://github.com/rom1504/clip-retrieval

It does not appear to actually show you how images were used to train a generative AI.

Also Ubuntu:

    > Subscription  Quantity  Billing        Total
    > Ubuntu Pro    1         Billed Yearly  $500.00

https://ubuntu.com/pro/subscribe

This is greater customer base suicide than the recent Netflix policies. "Pay up or get hacked".

SD2's dataset was deduplicated prior to training. That's why the paper is about SD1, which was a prototype model completed before Stability even had VC raises.

Technically we also trust Signal not to push a signed update to their software that exfiltrates our keys. Whether or not Meta is doing E2EE should be clear from snooping their protocol or reverse engineering the software.

$10k today, or in 40 years when that's how much I need to pay for a cheeseburger?

"Activist investors" engage in what will be considered a crime in the near future, which is purchasing shares to obtain board ownership -> take in obscene amounts of debt for stock buybacks -> dump shares and let the company fall into bankruptcy. I prefer their original name "corporate raiders".

When I order a cheeseburger and pay with cash, the clerk is always cautious to ensure that the cash I provide wasn't previously stolen. After enjoying my meal, I always make sure to thoroughly investigate the diner after to ensure that they pay their fair share of taxes.

Absolute bummer.

For anyone unfamiliar, Faithless was a huge group in the electronica scene in the 90s and early 2000s. Some well known hits include "God is a DJ", "Insomnia", and "We Come 1". If you're young and don't know them, check them out. For us older folks, lots of early morning memories of better times. :)

I am pretty amazed by how good mastodon actually is. It's like Twitter without the noise. The website defaults to chronological order and very few people assault others with crazy radical comments, because there are no rage-promoting attention-grabbing algorithms in place.