CuriousCosmic

That's not entirely accurate. Telomeres shorten with age but cancers generally are triggered by mutations or defects much further in the DNA away from the telomeres.

So one hypothesis is that telomeres shrink with age which causes runaway growth (i.e. cancer) to burn up all it's telomeres early and start crunching through sequences responsible for building proteins essential to life, therefore causing some of the cancer cells to die out. Whether that hypothesis holds is a different matter but the referenced paper points towards various other mechanisms which are seemingly caused by cell age as well.

Yeah they mention this in the weaknesses section.

> While phi-4 demonstrates relatively strong performance in answering questions and performing reasoning tasks, it is less proficient at rigorously following detailed instructions, particularly those involving specific formatting requirements.

Gonna need a source on that one.

As much as I despise Elon Musk, the cost per kg to LEO has gone down a lot in recent years thanks to spacex.

Statistically most would be. Assad's regime were Shiites (~13% of the population) and the HTS are Sunni (74% of the population).

This doesn't necessarily mean they are better but at the very least of two very theistic groups, the latter better represents the population than the former.

And of course while this means Russia and Assad are finally out of the equation, the governments in North and East Syria aren't exactly stoked about the HTS' explosive growth in power.

I was linking the proposers day video not as a direct answer but rather to try and show how that specific aspect is connected to all the other research going on on this topic as well.

And they even mentioned in that video that TA2 is the particularly hard problem.

And to address your question I linked the teaming page which does go into what the teams are pursuing to solve that problem. One of the teams addressing the specific issue you brought up is the USC team. They are developing electric field stimulus systems to drive regeneration of optic nerves along with real time imaging/monitoring equipment so that they can control said stimulus to drive regeneration along the paths they want.

TA2 is very big/broad and a lot of different teams are taking quite different approaches at how to address aspects of TA2.

Yep. I worded it a bit awkwardly but that was what I was getting at. And mostly I was trying to make the distinction between damage within the "eyeball" vs more comprehensive damage further up the optic nerve as that distinction would determine how viable a transplant would be at recovering lost vision.

> Their site force redirects to a relatively useless country-local top-page so I have no idea what TFA is about.

FWIW there is no article in this case. TFA is the shop itself, presumably to start a discussion of biohacking, particularly some of the implants they have.

So the short TLDR answer is that they are only working on one part of this. There are 30+ teams who were awarded funding and are working on the THEA project across the US [1]. Those teams are each working on aspects of one of the three different technical areas. Those technical areas are:

TA1: Retrieval of donor eyes and tissue preservation

TA2: Optic nerve repair and regeneration

TA3: Surgical procedures, post-operative care, and functional assessment

So this is very much a divide and conquer problem and no one team is planning to just figure out the whole solution. They are all still very ambitious projects but they are targeting much more discrete problems and working with other relevant teams on the project to bridge the gaps between their specific projects. To get a better idea of how it all ties together you probably want to skim through their proposers day presentation [2].

1. https://arpa-h.gov/research-and-funding/programs/thea/teamin...

2. https://www.youtube.com/watch?v=fRNpeU8_RLo

These are all different projects awarded by ARPA-H as part of their THEA project [1]. They are all aimed at the same end goal but they have different approaches and specific focuses. Generally with ARPA and DARPA projects the best way to get a broad scope of what all is going on is via the proposers day videos [2]. And of course take a look at their teaming page to see what all 30-40 of the sub-projects/teams working on this overall project are doing [3].

1. https://arpa-h.gov/research-and-funding/programs/thea

2. https://www.youtube.com/watch?v=fRNpeU8_RLo

3. https://arpa-h.gov/research-and-funding/programs/thea/teamin...

So it's a bit of an open question exactly what causes glaucoma. Optic nerve degeneration is one of the potential causes but nerve damage within the eye due to decreased blood flow or due to increased intraocular pressure are both possible.

In the latter two cases, basically all of that damage is limited to within the nerves within the eye and the optic nerve head whereas more general degeneration of the optic nerve would likely occur across the entire optic nerve up to and potentially through and past the optic chiasma.

This project (THEA) is primarily about pursuing whole functional eye (i.e. the entire eye "ball") transplants and reconnecting/regenerating the optic nerve to a new eye. If this succeeds it would mean that any degeneration in the retina or optic nerve head would be curable. And potentially the techniques pioneered in regenerating the optic nerve would provide a starting point for addressing degeneration further up the optic nerve all the way through the optic chiasma and optic track up into the brain.

So TLDR: If this project is successful it should mean a functional cure for most degenerative eye diseases even including most forms of glaucoma. And potentially it could lead to inroads in curing dementia related vision loss and other cognitive decline related vision loss.

https://arpa-h.gov/research-and-funding/programs/thea

The colliquial "x86 License" is the AMD-Intel Patent cross-licensing agreement. i.e. All patents related to x86 or any extensions of the ISA are automatically cross-licensed between the companies. While in the past the ISA patent story mostly was leaning in Intel's favor, since AMD64/x86_64 really took off, ISA innovation really became a delicate stack of cards interwoven between Intel and AMD.

So if Intel sells, everyone is fucked until whoever buys can renegotiate the terms of the agreement. And if that's Nvidia, they could just sit dead on the IP and starve AMD of the bulk of their CPU revenue (which is what is keeping the company solvent). And requiring they keep the agreement they currently have would mean requiring AMD to give Nvidia a pretty detailed look at the AMD secret sauce which would increasingly push AMD into the red until they become insolvent, again leading to a Nvidia monopoly.

The US government as well as the EU will not allow that to happen so however things slice, ownership of the x86 ISA patents would not be going to Nvidia.

Unfortunately rights to the x86_64 license expire on the event of transfer of either company to a new owner.

Also noting that you can occassionally get a "tortiose shell" esque male cat via a pretty rare process known as chimerism. i.e. two or more different embryos merge in the womb resulting in different parts of the body being formed from different embryos (meaning different DNA depending what part of the body you sample from). And of course for this to result in a cat you can see, they have to get lucky enough that the combination of embryos still produces a viable fetus.

Hey I mean two negatives make a positive right? Can't possibly go any worse than it already is.

What do they do that Nvidia doesn't (and that Nvidia would care about)?

They already do networking, photonics, GPUs, high speed interconnects, and CPUs. They are planning on selling their FPGAs (the Altera acquisition) to Lattice.

The only things left are their fab ops, thunderbolt/usbc, wifi, and ble.

Their fab ops would take over a decade of heavy investment to catch up to TSMC or Samsung and idk if even Nvidia is ambitious enough to take that on.

Wifi and BLE could be good additions if they wanted to branch out their mellanox portfolio to wireless. Thunderbolt/USB C also might be worthwhile.

But that IP is probably going to be cheaper to buy piecemeal so idk if it's worth it to buy the whole company outright.

Maybe they'll sell Intel to Northrop Grumman /hj

If you want an electric truck (or potentially an SUV), consider looking at an Edison Motors pickup truck retrofit. They are technically Diesel Electric instead of pure electric but you can customize the battery load if you want to run full electric. They don't do all the stupid cloud connected software stuff and they are all about repairability/self maintenance.

Probably the only electric vehicle manufacturer that isn't egregiously tech-bro-y and dripping in dark patterns.

1. Use a service like Bluesky Follower Bridge to map over your followings from Twitter to Bluesky (it's not perfect so you need to check each follow to make sure it didn't grab the wrong person)

2. Seed your account with "Starter Packs" that bulk follow users with specific niches. You can search for these on bsky. Like there's one for C++ developers, a bunch for artists, some for journalists for specific niches, etc.

3. Customise the bluesky moderation service labeller.

4. Subscribe to some additional labellers.

- A major one is https://bsky.app/profile/skywatch.blue. The account also has a number of moderation lists you can use to auto-hide or auto-block problematic accounts.

- If you care about US politics this labeller is useful: https://bsky.app/profile/us-gov-funding.bsky.social

5. Adjust your moderation and threading settings. Try out the new threaded replies mode. Try changing the reply ordering settings.

6. When you use the "Discover" tab, click the triple dots (...) on a post and click "Show more like this" and "Show less like this" to tune the algorithm for the discover tab. It starts out really bland/boring/random but it tunes in pretty quickly if you work with it.

7. Browse the Feeds tab to find custom feeds you like for specific niches. Add those to your home page so you can switch to them easily. These feeds will be generally a lot more focused than your main feeds but they are super useful for only seeing specific content you care about.

Replies aren't necessarily ranked based on merit btw. They are ranked by whatever ordering you choose in your settings. The options are:

- Hot Replies first

- Oldest Replies first

- Newest Replies first

- Most-liked Replies first

- Random (aka "Poster's Roulette")

I believe "Hot Replies" first is the default but I may be mistaken.

This seems to be where they are getting the data from (as they get their data from FinancialTimes who get it from SimilarWeb)

https://www.similarweb.com/blog/insights/social-media-news/b...

Specifically what it looks like they are pursuing is an SoC built around an FPGA that is optimised for runtime reconfiguration speed and presumably with an ISA extension to request and then later release configurations.

Wait I wasn't aware this was a product in the first place but isn't it really kind of scummy to try and namesquat on a famous piece of software that's well over 50 years old and available on practically every computer or device in operation?

Mystery Flesh Pit Beach?

Well technically anything can be a shuttle because specifically the thing that makes it a shuttle is the operating pattern (repeated point to point service) rather than the machine itself.

Etymology-wise a shuttle was a type of weaving tool which is why the verb shuttle exists, i.e. to rapidly move back and forth across a length (as if you were weaving a thread into a piece of fabric).

So then you got shuttle trains which frequently ran back and forth. And from there other types of shuttle services (shuttle buses, shuttle vans, etc).

And of course eventually the space shuttle being intended to be a launch vehicle designed for shuttle service to and from orbit. (side note but technically if the SpaceX Starship actually achieves it's intended sub-24h turn around it'd be able to qualify as a shuttle provided it ran a fixed point to point route on a regular basis).

Previous discussion:

https://news.ycombinator.com/item?id=42160659

Sure it's open source but that doesn't mean the public repo is in any way up to date. Especially given that the repos only have a handful of commits from the few months after the repos were published and they haven't been updated in well over a year.

https://github.com/twitter/the-algorithm

https://github.com/twitter/the-algorithm-ml

Moderation lists and labellers honestly already get you most of the way there. Labellers are very effective at flagging spam/botted content and accounts that continuously show up on labellers as spam/bot content get referred to moderation lists dedicated to specific types of spam and bot content.

So you can already start by using a labeller and just hiding that content behind a warning (kind of like the NSFW wall), hiding it entirely, or just attaching a visual tag to it (based on preferences). And then to filter out more consistent perpetrators you can rely on mute/block lists.

> I have a feeling that this is going to create a weird thing of some magnitude where accounts end up on popular blacklists for poor reasons and have no recourse.

This actually has already kind of occurred with moderation lists and the solution has generally been to strike down list managers who abuse their authority and block them (as you can't normally add someone who has blocked you to their lists).

Oh and here I thought you were talking about Elements of Programming by Stepanov and McJones which tbh I'd give the same recommendation/review.

https://elementsofprogramming.com/

Android Auto works well for me on Graphene (Pixel 9 Pro) but I obv can't make GPay work given that it relies on the OS having Google's blessing and they are actively hostile to GOS as a project.