804 karma · joined July 10, 2013
1. Enormous screen bezel, which made it silly to lug around a laptop with the screen real estate of one much smaller.
2. Without careful, tedious manual power management in linux the cooling fans routinely spin up to dust buster sound levels under normal use, which is unacceptable if you exist around other human beings.
The Surface Book 2 is an interesting option, but the linux support, while improving through community effort, is still spotty [1]. At the moment it has many power draw, heat, and noise issues. Also, everyone I know who has had a surface device has had to have it serviced or RMA'd at least once, which is not acceptable for something so expensive from such a large vendor.
I only have minimal experience with it, since I've also eventually chosen actix-web for all my projects so far due to its simplicity.
The point of rust+wasm benchmarks is that one can write reasonable, maintainable, functionality-focused code (not to mention all the rust-specific benefits) and get good performance out of the box.
We are looking to hire a software engineer for at least one year to work on the implementation of automatic neuronal arbor reconstruction from volumes of electron microscopy: a central problem in the emerging field of large-scale cellular connectomics.
The work will be centered on bringing automation to the software CATMAID[0], in particular by applying flood-fill networks and their tensorflow implementation[1] to build a workflow that will speed up the work of dozens of laboratories already using CATMAID for computer-assisted yet manual reconstruction of neurons for mapping wiring diagrams with synaptic resolution.
The engineer will be embedded in the Cardona lab at Howard Hughes Medical Institute Janelia Research Campus[2], among computer scientists and neuroscientists working together to map the wiring diagram of complete brains at synaptic resolution.
Optionally, solve an open GitHub issue for CATMAID, submit a pull request, and include this with your CV and statement. Applications including a pull request for an issue will be given priority consideration.
Official listing and more information: https://hhmi-openhire.silkroad.com/epostings/index.cfm?fusea...
[0] https://github.com/catmaid/catmaid [1] https://github.com/aschampion/diluvian [2] https://www.janelia.org/lab/cardona-lab
e: sorry some connection interruption and now this is redundant to other comments.
Beyond these, the main reason is that gaming is a much better exploration and experimentation field than productivity because each app can have far less capital investment, lifecycles are short, and there's no need for a reliable, successful output.
Or, you know, understands and thinks in the semantics of the underlying tool and is already working in other text-based tools.
Hypothesis: these businesses all monetize # of interactions (even if CPI is low), which they want to maximize by structuring interactions such that there is constantly new content that encourages obsessive checking. A sense of ephemerality of content reinforces that behavior ("I have to check because there is new content and it may be gone if I wait to check").
Yet (modulo mobile) somehow companies managed to do this in the 90s with far less productive development tools. The difference is the diluted power of the consumer: when software was a purchased product made for Ks of subscribing customers, small contingents of customers being upset about subpar UI and performance was a real threat to the viability of your business. Now Outlook or any Amazon app can be absolute slow, buggy garbage on nearly every platform (and they are), but the enormous customer base (diluting collective action) and entrenchment of locked-in platforms and data mean there's little incentive to obsess about product quality. Instead companies prioritize cost reduction, new customer acquisition (focusing on product chrome refreshes instead of robustness), or just give over to development momentum apathetic to quality.
Because most "actual native apps" now are not native apps but electron or some other HTML + JS + CSS wrapper running an entire separate browser to render the app. By running the web page as separate windows you're effectively getting the same app but sharing the browser overhead.
We are nowhere near this point. There are specific, intentionally constrained behaviors in model animals such that we can monitor the activity of specific sets of neurons and make predictions (meaning anywhere from better-than-chance to perfect accuracy) of behavioral output. This is far from "mapping out behavior" for an entire organism.
As for the GPP about a comprehensive look-up mechanism from all possible states of dynamics in the human brain to semantic descriptions of mental states: the problem is that trying to implement such a mechanism within the physical and thus computational constraints of our world will result, with high likelihood, in something that functions rather like a brain.
Also makes it easier for external clients (CLI, etc.) to use the service as an API.
The claim this ended up in Frontiers because no one else in neuro theory has heard of algebraic topology is laughable.
e: To be clear, I don't think this rises to the level of "Sokal hoax" -- it looks like an interesting result. I also have an interest in the role massive simulation can play in discovery. But to misrepresent topological analysis as an entirely novel approach or package it with phys.org-press-release nonsense like "mapping the brain in 11 dimensions" is both a disservice to the rich field on which it's building and adds noise to public communication about neuroscience, which is already plagued with bullshit.
[1] https://www.amazon.com/Vagueness-Reader-Rosanna-Keefe/dp/026... [2] https://www.amazon.com/Vagueness-Problems-Philosophy-Timothy...
As an alternative, though, there are several Rust-based scripting languages that attempt to expose some of the power of the type system, etc., while being more amenable to dynamic loading, REPL, etc:
Common interaction cues that span functions but aren't in the CSS set: multiaxis, rotation, etc.
Not to mention the innumerable domain-specific possibilities. It's another example of the tension between the idea of browser-as-the-future-ubiquitous-platform and browser-as-lowest-common-denominator-mobile-safe-sandbox, which is eroding the established niche of desktop software for deep, focused, domain-optimized work because of the tyranny of CRUDy consumption-focused web apps.
We build web applications serving expert users for which this is essential. For precise selection the default cursor often occludes important information. Think of how confusing something like Photoshop, AutoCAD or Maya would be without contextual cursors.
It would be sensible, however, to place behind a permission in the same way as location or notification APIs. Unfortunately most browsers' UX for permissions is awful.
So, yes, we need to solve this with technology -- policy itself, technology for improving the creation of policy and the selection of those who turn it into legislation, technology for combating disinformation and educating the public on protecting their rights, technology for detecting and disseminating information on relationships between entrenched interests and policy makers. Networking technology of the kind you mention may be part of that, but on its own it can never be a panacea. Policy will always necessarily exist, and it will never be solved -- it's a dynamic system.
Except it's none of these things: electron apps suffer the same presentation-before-content problems as most of the web, they inherit all of the state bugs of web apps, and they're painfully slow despite usually just being menus of nested lists and text boxes. It's passing a burden from developers to users, which considering the huge asymmetry between these two groups that's typical (at least 3 orders of magnitude, and often 6 or more) is an enormous inefficiency.
I'm all for selecting tools and platforms to better optimize developer productivity, but it's not clear to me the web platform has actually accomplished that for all but a few types of applications, and pushing it to the desktop is resulting in cumbersome, unresponsive, needlessly fragmented interfaces.
Put simply: Amazon Music Desktop has untold development investment and a complete visual redesign every 6 months, but WinAmp circa 2001 is somehow a thoroughly better way of searching and playing music.