https://www.nature.com/articles/s41586-019-1907-7

When we first started getting good at sequencing the DNA of tumors, I remember initial reports of taking samples across the 3D space of a tumor and finding great spatial heterogeneity in the tumor genomes.

I'm actually most excited for using this drug in combination with colon cancer, where KRAS mutation is a common drug resistance evolution in response to drugs that target the gene EFGR (though cancer researchers may all have their favorites to go after, colon cancer went after my family especially hard).

However this is just the first version of the drug, it can be combined with other modalities to allow more selective targeting of cancer versus not cancer cells (e.g antibody-drug conjugation). And when used in earlier stage cancers, rather than the advanced cancers in this first clinical trial, there's the possibility of lower dosing that has less strong side effects.

This is just the first attack that has ever broken through to hit a key weakness of some cancers. It's the start of learning, a breakthrough that will launch refinements, enhancements, and a ton of innovation. That sort of innovation is sometimes derided as "me-too" drugs, and not meaningful, but some of the biggest advancements in cancer care have been from taking very hard to tolerate treatments and making them more tolerable and refined and better for patients, allowing longer and more thorough killing of cancer cells. I would expect we will see a lot of that here, as well as work towards combinations with other drugs.

This is a horrifying proposal not only on the ethics front but also in the scientific uselessness of it.

This is exactly the type of thing that gave the Nazis the bad name they deserve.

https://clinicaltrials.gov/search?intr=daraxonrasib&viewType...

The first two are the trial that just completed and showed success: people that have pancreatic cancer that failed other treatments, then a "trial" that is meant to give quick access to more people now that it's been shown to work.

Then there's a trial for using it as the first-line treatment for pancreatic cancer, one for lung cancer (NSCLC), and also various combinations with other drugs. I expect we'll see a ton of new trials registered in the coming year. Especially something in combination with colon cancer, because a common drug resistance mechanism in colon cancer is to develop KRAS mutation.

The thing is that we don't really know which cancers it will work well in until we try. And there's limited number of people with cancer that enter clinical trials, and we want to give each person their very best chance at survival, and then there's the massive expense of running the clinical trial itself, so learning happens slowly, one month of survival at a time, or one cancer recurrence at a time, or one death at a time. Patients that take part in clinical trials really are the heroes here. (Especially with the side effects of this new drug, which are horrible. It is a revolutionary drug, but we need to learn how to manage the other things it does as well, and that's going to take time.)

One of the most commonly observed broken mechanisms is mutation in the gene KRAS that turns this on/off growth switch into the permanently on position.

This has been known for decades, of course. And there have been huge amounts of effort to try to develop drugs that target KRAS in cancer, but for decades it's always been thought of as 'undruggable' because of the difficulty of finding any molecules that would affect it.

This new drug, that finally treats KRAS mutated cancers, goes about it in a new way. Instead of trying to gum up the works of a single protein by sticking a small chemical in it, it effectively "glues" the KRAS protein to another protein, CypA, which keeps the switch away from reaching the normal areas where it's "on switch" activity works.

So this new drug means two things: 1) a lot of the most difficult to treat cancers are now far more treatable, and in the next 1-5 years clinical trials will tell us which cancers this particular drug works well for, 2) there's an entire new class of drug activity that everybody is chasing at this very moment, so in 5-25 years we'll likely have a huge number more of these sorts of treatments.

The only problem with this comparison is the cost of the grid, which at least in the US will dwarf the savings from doing utility scale installs.

One nice thing about residential solar is that it greatly reduces the need for transmission, while perhaps requiring some short-term enhancement on distribution. And ideally dumping a bunch of storage at grid distribution nodes in the form of a container of batteries would solve a ton of problems and reduce costs a lot. This is of course heavily resisted by utilities because in most places in the US they make their money on T&D costs, so cutting those is a threat to their existence.

https://www.eia.gov/energyexplained/electricity/electricity-...

The 10-year change 2015 to 2025 is:

- Gas: +472 TWh, +35%

- Renewables: +525 TWh, +97%

Saying that this is more gas than coal is certainly not the case borne out by the numbers, even in the US, the one place where gas is as cheap as dirt due to it being a by-product of fracking.

Natural gas's share of electricity generation has been falling for five years straight:

https://ember-energy.org/latest-insights/gas-share-in-global...

https://www.eia.gov/todayinenergy/detail.php?id=67205

- Solar: +87 TWh/year (assuming 23% capacity factor, lower end of US range)

- Gas: +9TWh/year (6.3GW new, 4.6GW retirements, higher end of US capacity factor of 60%) https://www.eia.gov/todayinenergy/detail.php?id=67206

This is in the face of massive growth for grid demand for the first time in decades, so the trend will accelerate.

New gas turbine manufacturing capacity is tapped out, causing new gas CapEx to get more expensive:

https://www.woodmac.com/press-releases/gas-turbine-prices-so...

Meanwhile solar and storage are continually plummeting in price.

So the current trend of approximately all new generation being renewables is going to accelerate. And then it will start eating into older, existing generation assets, causing early retirements of existing gas generation capacity.

Most investors think that any new gas generation built today will be a stranded asset long before its end of life. That doesn't matter to the hyperscalers, who run them so poorly and hard that the turbine shafts die in a few years and can afford it, but for regular utilities, buying any new gas generation is a boondoggle meant to soak the ratepayers and capture the guaranteed profit rate.

And the numbers above ignore residential solar, which will further lessen demand for gas, and as the cost of transmission and distribution soar on the grid, residential solar becomes an always better deal, because it skips all that.

The global cost-minimum for a future grid will have gas on it for maybe 20 more years, but not much after that. We'll switch to lots of storage and tons of over-capacity of solar and wind.

However I think your EV examples shows an important attitude about what types of vehicles can be regulated. EVs are fair game for regulation, oversize trucks and SUVs are not. That's an attitude not based on safety, but on societal priorities.

This two-class system extends even beyond safety regulations, into emissions regulations too. Trucks and oversize SUVs get a free-ride out of everybody else in society.

41% of vehicle deaths are people not even in a car[1]. Yet car safety regulation is heavily focused on the 59% that are, nothing to regulate the ridiculous gender-affirming hood heights or aftermarket lifts that turn a survivable collision into a deadly collision.

[1] https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/... Table 1, paragraph above

Comments URL: https://news.ycombinator.com/item?id=48477415

Points: 2

# Comments: 0

I think Section 9, and Figures 13/14 in the Arxiv preprint sort of address this, but it doesn't mention anything about accounting for real-world failures in fat trees. I haven't had a chance to read it all, though...

> The results were striking: compared to traditional fat-tree networks, RNG (Resilient Network Graphs) uses 69% fewer routers, delivers 33% higher throughput, cuts network power by 40%, and lowers operating costs by 27. In early 2026, RNG became the default design for most newly built Amazon data centers globally.

> For cabling, they developed the ShuffleBox—a passive optical device whose internal wiring combined with randomized ShuffleBox-to-ShuffleBox cabling yields “quasi-random” graphs that behave like truly random graphs.

This is pretty incredible, random layouts of networks that have on-average better properties...

I'm really curious about the long tail of performance though. What is the worst case scenario here? And are there some better case scenarios? Uniformity in Clos networks is pretty great, but many loads don't need uniformity, and if these RNG-based networks have non-uniformity, perhaps that has operational characteristics that can be helpful or harmful.

The rest of the world wants dollars, and as long as they do, the US benefits massively from that demand, and it would be foolish not to satisfy that demand, as it's pure benefit to the US.

It seems that many are devoted to ending the status of the dollar as the reserve currency of the world, but it's a strange and subversive thing for any US resident to want or desire.

I don't think you can call the process unrelated to the mother or the baby, they're both pretty important throughout the whole thing.