Hacker Newsnew | past | comments | ask | show | jobs | submit | Teknomadix's commentslogin

Highly editorialized title. "Transient multidomain functional improvement in advanced Alzheimer’s disease following high-dose psilocybin-containing mushroom administration: a case report" This is exactly that. A case report. This is not a successful treatment. Far short of that.

> We report the case of an octogenarian Japanese-American woman with a 10-year history of Alzheimer’s disease, including 5 years of marked hypofunction and predominantly monosyllabic speech.

> The patient received 5 g of orally administered psilocybin-containing mushrooms

> Approximately 19 h post-administration, spontaneous autobiographical speech emerged. Over subsequent days and weeks, functional improvements included restoration of urinary continence, improved ambulation, autonomous dressing, increased emotional responsiveness, sustained social interaction, contextual memory retrieval, preserved working memory for social context, and spontaneous conversational engagement.

This is exactly that, a treatment using psilocybin that was successful. It's not claiming to have developed a treatment protocol, the title is precise.


The title is misleading. “Improvement in advanced Alzheimer’s disease following high-dose psilocybin” sounds like results from a broader study, not an anecdotal report.

There’s a reason the article includes “a case report” in the title.


It might be inappropriate for the target audience because they might mistake the treatment for the protocol, but it's not trying to be misleading I think. In any case the title got editorialised which I think settles this.

It appears that one dose of the shrooms had positive effects observable for several weeks. That is very encouraging and hints at this being a practical treatment. (The word "transient" in the original title sets us up for disappointment, conjuring up the image of an experiment in which Alzheimer's symptoms are suspended for 20 minutes).

Questions linger about the long term use; what happens to someone who has been taking this dose, say, every two weeks for a year, or five years? Does even the second dose work as well as the first?


So is this about temporarily overclocking some cells that would then return to a healthy baseline before collapsing back after the overclocking agent is gone? Why not just do 500mg B1 HCl IV then for a comparison to see if this was just a metabolic block?

Cool idea. Horrible controls. Reminds me of how frustrating playing pacman on my 2600 was as a kid.

    The code was written by Claude, unfortunately, and hence no controls were probably even considered, or no tokens were left.
    I jumped in with the love in mind, too, but when I checked the source repository, and saw the actual source... and then the contributors... it was... I am sorry... it hit hard...

Where did you quote this from?

> Horrible controls.

Interesting! I found them mostly ok, but I was playing with a keyboard. They definitely remind me of the controls of Pacman games I grew up with where (as sibling comment notes), you have to decide where you're going before you get there.

I wonder if tweaking the input buffering or adding some frames after the turn where you 'snap-back' would help, similar to ghost jumps in Mario.


I grew up playing Pac-Man and Ms. Pac-Man in dimly lit arcade parlors and bowling alleys.

The controls in this version are, for lack of a better word, sluggish compared to the tight responsiveness of the originals on a four-way joystick or using a keyboard with MAME. Even when you press an arrow key to "move" the ghost, there’s a noticeable delay, almost like it’s polling for the key-up event instead of the key-down.


For me there seems to be latency registering the key press which results in frequently missing turns

I find that you have to decide which direction to turn before the turn.

Once I got that figured out, it was mostly a race to catch pacman along the long straight path at the bottom edge with the ghost's superior speed.


Swipe controls on mobile worked well for me.

Yeah its AI slop without the corner easing. But maybe they can AI it better.

CAD / CAM & Machining. Co-bots help tremendously at the present, and once a job is 'running' it's very automateable, but getting a job to that point of automation takes tremendous human skill and efforts. Machining touches every physical thing around you. There exists huge potential for more machine learning and automation with coming generations of robotics and edge computation, yet we are a long ways from generalized robots and embodied "AI" that are be able to conceive a new part or whole machine from its parts and make it out of mill stock. LLMs alone will never accomplish that frontier.

So you do two websites over the course of a week. You must do cold calling and outreach to maintain that momentum. You are not making $1k/hr. You have a ~$50/hr side hustle before you've paid your ±35-45% taxes.

that's not terrible in many places. high chance the OP is not in the US

US Govt reduces Fable Tokens by 100%.

Space lasers.

Isn't there a better alternative to this archive site?

What's wrong with it? It shows the article, with a little banner on top, showing when it was captured. Doesn't mess with scrolling, what's not to like?


didnt the archive owner only start doing this after Patokallio revealed his identity for no good reason. Given the legal liability involved in maintaining that service, there is a threat to what Patokallio is doing so it doesn't seem entirely unsympathetic to do something in retaliation..

Seems like a mistake to rely on an archiving site that runs malware and changes the text of archived pages.

However, there’s also not really anything better out there. The owner being some wily/sketchy Russian is even an advantage in some respects, since they can’t be pressured by copyright law or foreign politics.


Hrrm :-(

Well, for one, it wants me to scan a QR code as a captcha.

That is an issue you can pin on Google:

<https://reclaimthenet.org/google-broke-recaptcha-for-de-goog...> (8 May 2026).

HN discussion: <https://news.ycombinator.com/item?id=48067119>

Though yes, Archive.is have continued using reCaptcha despite this recent change.


I've never once had that happen. It's always straight to the unrolled article.


I chose the audio option to get around this.

Me too, and ain't no way I'm doing that. That's blocklist behavior.

Why does everyone on HN complain and post about anything and everything except the content of the link.

It does the job for me, but please do tell me if you find a better alternative.

These are interesting, and offer beefy through put. No point in adapting to a PCI lane thought, stuck behind the slot-bus bottleneck.


Tesla V100 SXM2 16GB is NOT DGX class as the author writes. It's HGX class. The V100 comes in two classes, SXM2 and SXM4, the latter coming with a Max of 80gb on board memory. Typically these are installed 8×A100 80GB SXM4 on an HGX riser, and what that gives you is NVSwitch fabric and 640GB of pooled HBM2e (on package stacked memory /w ~2 TB/s of memory bandwidth). 2u standard rack footprint too.


What on earth are you talking about? Your comment makes no sense.

The V100 and A100 are different generations altogether.

The V100 does not have 2TB/s.


I have no idea what you are trying to say.

V100 came as sxm2 and sxm3. And it was 16 and 32gb.

HGX is DGX with extra toppings.


No. DGX is when it's Nvidia's design, HGX when it's a 3rd party design.


Stoned punks for sure.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: