> 100× energy-efficiency advantage for running some of the largest current Transformer models, and that if both the models and the optical hardware are scaled to the quadrillion-parameter regime, optical computers could have a >8,000×
Maybe I interpreted that incorrectly but I thought it's saying a 100x advantage for current large Transformer models, and 8000x advantage for future quadrillion-parameter models? I didn't include those because I suppose that size of model is quite a few years away. Admittedly this is only based on the abstract...
> it may be possible to achieve a 100× energy-efficiency advantage
Running the math on a machine with 8x A100 (enough to run today's LLMs), that would be 300w * 8gpus / 100 = 24w.
This is within striking distance of IOT and personal devices. I'm trying to imagine what a world would look like where generative text models are commodetised to the point where you can either generate text locally on your phone, or generate GBs of text in the cloud.
I have to admit it's very hard to make any sort of accurate prediction.
From the info in the article alone, it's impossible to tell if this is even a correlation at all. I don't know what the expected number of reports would be for a given year, and if 30k is higher or lower than that.
> If a link between vaccination and menstrual changes is confirmed
Even the wording in the article stops short of claiming there is evidence for any link. It's just a bad headline.
Reddit was funded by ycombinator in summer ‘05, which probably explains part of it. Then the rest I would say is down to similarities between the two sites (primarily acting as link aggregators, although reddit has gradually diverged from this).
Has anyone heard of good resources that teach you "when to use X" instead of "why you should use X"?
It's somewhat understandable that the official websites of frameworks/tools/libraries/languages/services tend towards espousing the advantages of the thing and leaving out most of the drawbacks. That's just how the incentives are aligned. But even the average blog article about a tool posted on HN rarely talks about what situations it is most effective, and what situations it should potentially be avoided. I suppose it's a natural consequence of blog authors only knowing so many tools, no single person knows them all.
I'm not saying this information isn't out there – just that it seems like the exception rather than the rule.
I want that protocolization too, although I don't hold out much hope that the monopolies in place can be broken, outside of fairly radical regulation.
To add to your list, there is also Secure Scuttlebutt [1] which has had a decent userbase over the past few years, and Planetary [2] which is a funded iOS client for it.
I think in general they all suffer from the chicken-egg problem and will need some reason for enough people to switch to be able to build a userbase. There isn't really any "novel hook" like tiktok, twitter, whatsapp, instagram, snapchat, etc have had in the past.
I've been doing more of a product-ish role in my dayjob, so it's nice to do something (relatively) well-defined, where it's easy to tell if things are working, but still challenging enough to be fun.
The technique you're looking for is Acoustic Fingerprinting [1]. There's a pretty easily understandable python implementation[2] of it if you're interested. In Shazam's case it is looking for peaks in the sound's spectrograph, so I don't think it would likely confuse two songs based on melody alone. It would have to be very close in rhythm, melody, and possibly even be in the same key.
> It is hardly surprising that techies are not getting enough sleep, given the industry’s culture of long hours, and the widespread notion that for a true entrepreneur, everything else in life is secondary to succeeding at work.
This article was very snarky, even for the economist. It seems distinctly below their usual standard of quality with broad generalisations such as the above, seemingly contradicted by an anecdote about Bryan Johnson and others' bizarre sleep practices.
For me this article would have been a lot better if it focused on the sleep tech industry, and applied a more measured and evidence-based approach to its silicon valley criticisms, some of which certainly are relevant and worthy of discussion.
It's kind of hard to predict which way the fare would go for such a service... It might actually end up being higher than what Uber is currently charging. Which wouldn't be a bad thing necessarily.
One thing I've noticed since moving to a city where driving is unnecessary, is how good it feels to drive occasionally. While I think a big part of that is the physical sensation, I also think doing these types of "menial tasks" can put your mind into a mode that is not possible when being entertained by a smartphone/kindle/whatever.
Currently on my 35 minute commute, I can be using a smartphone about 90% of that time. Does that sound efficient? Sure. But I think the article hits upon a real point, if we're going to "save time" on a particular activity, it had better be a worse activity than what we end up replacing it with.
I think the problem isn't the % of population that lives in a particular hemisphere, it's that the seasonal versioning scheme forces you to think about which hemisphere the software was created in. I'd say most users of an open source database would be aware that there is software being written in South Africa, Australia, New Zealand etc.
I use IdeaVim at my day job and can confirm it's gotten better. There are still some rough edges, especially around handling movements like L, M, and H which don't always go where you expect them to. And some operations like undo don't play well with the vim commands such as undo.
And yes, I completely agree about switching between vim and ideavim for different languages, the experience is very different and the sight differences are very noticeable.
Maybe I interpreted that incorrectly but I thought it's saying a 100x advantage for current large Transformer models, and 8000x advantage for future quadrillion-parameter models? I didn't include those because I suppose that size of model is quite a few years away. Admittedly this is only based on the abstract...