A focus on the common properties of cancer would perhaps provide more hope. The most obvious is the need for angiogenesis: the formation of new blood vessels to irrigate deep layers of a tumor. There are drugs out there targeting angiogenesis. Another obvious property is cell proliferation, without which a cancer is not a cancer. Both of these are core functions of most of the human body, but there is such thing as therapeutic windows: a time interval with a dose just so to have a greater effect on tumors rather than on the rest of the body.
We are hiring a PhD student to develop algorithms for reconstructing neural circuits from series of electron microscopy images. The approach includes computer vision and machine learning techniques, and composing formal models of neuronal arborizations for a top-down approach to image segmentation.
Look up for example what did Nokia sell before selling cell phones. Or look into what PG says about his recruits: that he chooses the individuals, not the specific application proposal, and that generally the focus of the company is completely different by the end of the YC school.
1. Establish a comment review system: two steps to get the comment public.
2. Show a random subset of the comments to users under a threshold score. Would avoid to an undetermined extent the accumulation of votes in specific comments. Would help avoid the current winner-take-all problem in comments (the top comment is read more and gets more upvotes regardless of quality).
"But if you're wondering if you could squeeze a little extra performance out of your program by switching to QuartzGL — probably not unless the key limitation in your program is raw pixel fill rate and you have relatively few lines."
Perhaps unsurprisingly, fill rate is what improves the most when drawing on a java.awt.image.VolatileImage .
Very nice, thanks for sharing! For my own purposes I implemented the function in clojure, which I find more readable. Destructuring and literal vectors are great assets of clojure. Autopromotion to avoid overflow (as in common lisp) is also great.
For "DNA research" (or rather, searching for peptide sequence matches), there's HMMER at http://hmmer.janelia.org/ , which uses hidden markov models. Claims to be great at detecting distant homologs. And after an enormous engineering effort, it's as fast as BLAST (in practical use it's faster, because its servers are currently not known as widely).
A popular Soviet SF writer Sever Gansovsky in his short story "The Host of Bay" (1962) described a sea monster consisting of microscopical particles which can self-organize themselves into a unity for hunt.
There's a sea colony animal--siphonophore--that can change shape from a broad, amorphous colony mass into a streamlined compact cylinder, for swimming. The colony members are not free-living; the colony develops by budding. If the members were separate and assembled later on, science fiction would have been beaten again by a living organism. Does anyone know of an example of a species in which separated individuals assemble into a colony? There are descriptions of sponges, where dissociating the sponge cells, passing them through a sieve, and then letting it sit for a bit results in the cells reassembling. But the sponge doesn't look menacing neither predates.
"A functional, object-oriented programming language based on Java and .NET, Scala includes some constructs for parallelisation from the outset."
"The goal is to overcome the challenges of parallel programming, [...]"
I hope they copy clojure to death. Clojure got parallel programming constructs right--the first language to make parallel programming trivial (think of pmap).
(AFAIK clojure's PersistentMap has already been ported to scala.)
I has example code and comments. Just like http://php.net did and which gave it enormous popularity and ease of use. What the docs don't explain a comment or an example clear it up.
The clojuredocs author knows the site needs to become even better. I thank him for what the site already delivers.
Meta: interesting that you think HN will exist in a recognizable form by 2028. 18 years ago, the web looked very different. I wonder what will we call "web" in 2028.