453 karma · joined December 11, 2015
> Mice whose senescent cells were killed off over six months were healthier, in several ways, than a control group of transgenic mice in which these cells were allowed to build up. Their kidneys worked better and their hearts were more resilient to stress, they tended to explore their cages more and they developed cancers at a later age. Eliminating senescent cells also extended the lifespans of the mice by 20–30%, Baker and van Deursen report in Nature on 3 February.
> As for the “excessive synaptic pruning” model, it’s very far from new: it was proposed in 1983 and has since been discussed in 3580 papers. So it’s simply untrue to imply that Sekar et al. are the first to shine light into a biological “black box”
Yeah, people have been running their mouths for a century proposing hundreds of radically different schizophrenia theories. It's different to propose yet another one, and to show that one out of the hundreds is genuinely part of schizophrenia!
Contrarianism for the sake of contrarianism.
> ...technological civilization doesn’t last long enough for it to happen.”
Writer:
> Both York and Teller seemed to think Fermi was questioning the feasibility of interstellar travel—nobody thought he was questioning the possible existence of extraterrestrial civilizations
What happens when a technological civilization 'doesn't last long enough'? It stops existing.
It would have been much better to cite the SMPY longitudinal results for a meaningful idea of how child prodigies fare.
There's a lot of similarity. Take math theorem proving, which is an active area of machine learning; the algorithms tend at the moment to look something like create a tree of possible theorems, then the algorithm tries to traverse down the tree towards the target theorem by evaluating each branch by the probability it leads towards the target. Because of the exponential explosion, it's very important to select good theorems to expand at each level of the tree, and the ML algorithms can learn characteristics of good intermediate theorems by learning on the existing corpuses of machine-checkable proofs, which gives them much higher success rates in reaching target theorems. So right now the best theorem-provers can already reprove a lot of existing proofs. But they're using stuff like SVMs last I saw, and were much less sophisticated than a deep q-learner.
Some links: "Theorem Proving in Large Formal Mathematics as an Emerging AI Field", Urban & Vyskocil 2012; "Mathematics in the Age of the Turing Machine", Hales 2014 http://arxiv.org/pdf/1302.2898v1.pdf ; "Josef Urban on Machine Learning and Automated Reasoning" http://intelligence.org/2013/12/21/josef-urban-on-machine-le...
The computation requirements for training and running are given in the paper in both the body and the appendix details.
But this isn't apple and oranges; a blog with themes and whatnot is much more sophisticated than the HTML CSS rendering defaults. IMO, a HTML page without any CSS is not that bad looking and you don't really need it. Heck, even a text file written with a little bit of care is perfectly readable.)
And the information overload isn't much of a problem: most people know immediately that static site generators aren't for them.
> Look at medium, and look at instagram. A slightly better publishing experience dominates.
Look at all the blogs using Wordpress, and everything written on Twitter and Facebook.
> And to further that note, how would you self host a photo site for mobile? Is that even possible? How would you begin?
I am not entirely sure what a 'photo site for mobile' is, but I am sure that it has no equivalents from before 2006 or so which you could claim it is harder or more complicated than, and I suspect that you could get 95% of the value from a static HTML page sitting on S3 containing a bunch of <img> tags and a 'viewport' <meta> tag that you copy-pasted from a StackOverflow post you found by googling 'how to make a mobile html page'.
Let's compare modern blogging to how you would have blogged in 1996, or when Berners-Lee launched the first documents at CERN, your workflow looked like 'write .html file, FTP to server, Apache serves static files to anyone who visits'. In 2016, your workflow, if you want to, looks like... 'write .html file, SCP to server, nginx serves static files to anyone who visits'. You can still write HTML5 by hand, it's not harder than original HTML (may be easier). You can write Markdown which compiles to HTML with no more fuss than 'markdown.pl foo.md foo.html' or 'pandoc foo.md -o foo.html'.
You don't want to run a server? That's fine, find a host like S3 and instead of 'scp', 's3cmd'. It'll cost you next to nothing compared to back then.
You want some sort of blog software? There must be a million which are not named 'Wordpress' and do not have endless security nightmares. You want static site generators? Likewise, millions. WSIYWG HTML editors? Likewise. Free hosting? At this point almost any large site offers blog-like functionality because hosting is so cheap.
But what is harder? If you want to run your own server, everything is infinitely easy than in 1996. Apt-get may or may not have existed in 1996, but I'm sure it was far less comprehensive and reliable than it is now. You can let your server run for many years without upgrading if you want, and when you do upgrade, it's a few commands. Why, you don't even need to compile a kernel or buy a server edition of your OS. It has never been easier nor cheaper to run your own website in a myriad of different methods. So what exactly has gotten 'so damn complicated'?
The reality is that most people, and especially casuals, do not care that much about high-quality publishing or control or reliability, and instead care more about extreme convenience and network effects, and often either have nothing to say or don't want to say it and so no matter how polished your static site generator is, they couldn't care less. As Yogi Berra said, "if people don't want to go to the ballpark, how are you going stop'em?"
So the real benefits here are: works in mountains, and future-proofs against better GPS jamming.
They wouldn't. You're building miles of thick stone walls exposed to the elements, which you are deliberately forcing plants to cover as much of them as possible. The article and the followup hint at the repair and labor costs (note that the article is set in China):
> The decline of the European fruit wall started in the late nineteenth century. Maintaining a fruit wall was a labour-intensive work that required a lot of craftsmanship in pruning, thinning, removing leaves, etcetera. The extension of the railways favoured the import of produce from the south, which was less labour-intensive and thus cheaper to produce. Artificially heated glasshouses could also produce similar or larger yields with much less skilled labour involved.
This should not be surprising, since if fruit-walls were strictly superior, why were they ever obsolesced? In general, when reading sustainability fetish material like this, you should remember that these sorts of things are like code-golfing or min-maxing: schemes which place unrealistic value on just one dimension, and which can be highly interesting in how one maximizes that dimension, but are only rarely optimal under a more realistic set of tradeoffs. In this case, the fetish is for energy use while running, and mostly ignoring capital, repair, space, and labor costs.