HNHacker News
TopNewBestAskShowJobs

andbberger

1,774 karma · joined January 2, 2014

submissionscomments
andbberger··on Affordable housing in California now routinely tops $1M per apartment to build
sprinklers are not an incremental safety benefit. they're probably the single most effective bit of fire safety mandated by codes. mandates to install them in single family homes is overkill though.

sprinklers are so effective that single means of egress + sprinklers should be the new building standard

andbberger··on Affordable housing in California now routinely tops $1M per apartment to build
you're thinking of the second ave subway. that's where the $2B/mile figure comes from. and that's nothing compared to what we're about to blow on the gateway project. portal bridge was coming in at $10B/mile last time I checked, making it by far the most expensive bridge ever built. and it's just a normal bridge.

but if VTA gets it's way and builds the SJ BART extension i'm sure they can put out some impressive numbers of their own (this extension should not be built at any price btw).

the insane cost of transit development and housing in the states are two sides of the same coin. to a first approximation they are caused by the same thing, and cause the same problems in our cities.

andbberger··on What Is LaMDA and What Does It Want?
jesus christ
andbberger··on Lilium achieves first main wing transition for all-electric aircraft [video]
the hover power density is insane, 2500W/kg
andbberger··on Cooling the tube – Engineering heat out of the Underground (2017)
hardly primitive, it's by far the most efficient form of regenerative braking
andbberger··on Cooling the tube – Engineering heat out of the Underground (2017)
piston effect already does a lot of ventilation work

the more elegant solution in use around in the world is to put the stations on hills

andbberger··on What’s the best lossless image format?
you should use NIfTI, which is the de facto standard for 3D medical imaging.

not compressed, but handles the voxel-image space transform which is far far more important and quite complicated.

you can slap gzip on top of it and most libraries/FIJI will transparently decompress

andbberger··on Tesla’s ‘phantom braking’ problem is getting worse
the players in the early days of nuclear power weren't criminally negligent. self-driving isn't going anywhere while the states are still car-obsessed wastelands (which is to say never). it's climate cope. instead of changing the urban fabric, build some li-on powered self-driving cars and convince ourselves we're doing the work while we continue to hurtle to our doom.
andbberger··on S.F. population fell 6.3%, most in nation, to lowest level since 2010
SF is attractive it's just stupidly expensive
andbberger··on DARPA wants to combine traditional sealift and airlift
it's... not that radical of an idea...
andbberger··on The next shock for homeowners: Surging property tax assessments
prop 13 might not be the root of all evil, but they're definitely bedfellows
andbberger··on The next shock for homeowners: Surging property tax assessments
land is finite, valuable and non-fungible. beanie babies can be mass produced and are fungible.

> I understand you pay to be in high value areas like near major transit and cities. But there ought to be rural low demand areas where people can live without being surfs all the time

enter stage left: land value tax

andbberger··on Ambitious Brain Recordings Create Unprecedented Portrait of Vision in Action
check out the olshausen paper I linked
andbberger··on Ambitious Brain Recordings Create Unprecedented Portrait of Vision in Action
https://en.wikipedia.org/wiki/Two-photon_excitation_microsco...

https://en.wikipedia.org/wiki/Fluorescence_imaging

https://en.wikipedia.org/wiki/Calcium_imaging

https://en.wikipedia.org/wiki/Genetically_encoded_voltage_in...

https://en.wikipedia.org/wiki/GCaMP

andbberger··on Ambitious Brain Recordings Create Unprecedented Portrait of Vision in Action
this was done with fluorescence imaging, not optogenetics, which is a perturbation technique not an imaging method.

i don't know where you're hearing that microelectrode arrays are due to revolution the field, but no neuroscientist thinks that. ephys is used for high temporal resolution activity measurements of a small population of neurons. it's not a new technique. if anything is revolutionizing the field it's fluorescent imaging which is in widespread use for imaging large populations of neurons (or even the whole brain in drosophila) at decent spatiotemporal resolution.

each technique has it's place.

> where there are practically purpose-built clusters of cells for certain features and contours

that's not really true. in mammalian visual cortex V1 is probably best characterized but we're still a ways from a complete understanding. see http://redwood.psych.cornell.edu/papers/V1-article.pdf

the optic lobe of drosophila has been characterized in far greater detail (where we afforded such luxuries as connectomes) but we are still a long way from a complete understanding of that

andbberger··on “Crime Pays but Botany Doesn’t” crosses citizen science with environmentalism
joe would fucking hate that he's on the front page of HN
andbberger··on Stanford gets $1.1B for new climate school from John Doerr
if you were in that position i would hope that you would do some napkin math about the necessary scale of carbon capture. last i checked the most efficient process would require about 1000x the annual energy output of the world just to keep up with emissions.

it's an understudied area, research funding has the potential to go a long ways. eg coccolitophores are sparsely studied but have the potential for carbon capture at a scale not achievable by industrial facilities

andbberger··on Deep Learning in Neuroimaging
carefully. that's the whole problem. the beauty of compressed sensing is that it comes with a convergence guarantee. figuring out how to provide similar (but probably not as strong guarantees) with more sophisticated models of the prior is what is going to allow this to work.
andbberger··on Deep Learning in Neuroimaging
imo the real value of ML in imaging is going to come from the use of variational inference in scope development. due to their raster scanned nature, off the shelf 2p scopes can deliver high resolutions or high framerates, but not both. 2p scopes using compressive sensing have been developed to get high resolution at high framerates have been developed, but there is a lot more room for optimization, and the scope i'm thinking of has convergence issues (compressive sensing generally has good problems but this one ends up being ill-posed).

variational inference is basically super-powered compressive sensing in this context. compressive sensing exploits the correlated nature of voxels in the iamge via some sparsity in some domain, it's a pretty weak prior. there's a lot more structure to be exploited. the statistics of natural tissue can be modeled with ML, as can the physics of the measurement process.

the trick is posing the problem in a way that comes with convergence guarantees. to my knowledge no one has done this yet. a lossy microscope is not very useful.

andbberger··on Costly SOFIA telescope faces termination after years of problems
not an astrophysicist so can't comment on the science directly. cost-benefit analysis in science funding is a really difficult problem. that being said, $85M buys quite a lot of science.
andbberger··on Costly SOFIA telescope faces termination after years of problems
results might not be significant, project might not pan out for any number of reasons, projects can easily take more than 6 months. sadly, people still don't care about negative results
andbberger··on Fruits and vegetables are less nutritious than they used to be
not like they had to reach for straws to write that story anyway - there is an interesting, nuanced narrative to be told on the haber-bosch processes influence on food prices and the practices of industrial agriculture, particularly tilling. we used to grow food in soil, now we grow it in dirt.
andbberger··on Ask HN: Teach me something new
QED is something you write after proofs not incomprehensible sentences. there is a compelling argument to be made that what neuroscience is missing is theory, not data. there is quite a lot of data.

i am dissappointed, although not surprised that someone who relies on appeals to authority when arguing a point is of the mindset that in an organism so finely optimized that it uses biomechanical feedback in gait generation where larger organisms use central pattern generations involving larger number of neurons that some neurons are more important than others. neurons are expensive.

guess you had better read some more papers.

andbberger··on Ask HN: Teach me something new
i really couldn't care less how many papers you've read. whole brain calcium imaging is routine, no one expects it to be the whole picture. it's useful though.

and no it is not a matter of debate, there are precisely 302 neurons

andbberger··on Ask HN: Teach me something new
indeed the focus of most research is the neuropils
andbberger··on Ask HN: Teach me something new
> we lack real-time datasets and we lack accurate instruments to observe and retroengineer the while system.

wrong again. whole brain 2p imaging is routinely performed in c elegans ( and drosophila for that matter) for a number of years now. in fact whole brain ephys in c elegans is feasible, although it would be an experimental tour de force. c elegans does not have '~150-300 neurons', it has precisely 302, and the complexity therein cannot be underestimated. it wasn't a layman explanation for I am not a layman, and you are missing my point, which is that reductive facts about such staggeringly complex things are neither interesting or useful. it's the feynman magnet explanation, but grumpier

andbberger··on Ask HN: Teach me something new
> The axons are just as important as the neurons.

what. an axon is a component of a neuron

are you referring to glia?

andbberger··on Ask HN: Teach me something new
> Like any brain in such circumstances (the hostile vacuum of space with no blood supply or body), it would almost immediately stop functioning and begin to deteriorate.

that is not such rare fluctuations are ephemeral. they are ephemeral because the dynamical system is reversible, for the same reason that the molecules in a box of gas dissipate immediately after they happen to coalesce in the corner

andbberger··on Ask HN: Teach me something new
this is a terrible list of 'brain facts'. they're not interesting or constructive to a larger understanding, and analogizing the cerebellum to a GPU is naive and misleading to the point of being insulting. working neuroscientist don't think about consciousness, we look at neurobiology, develop experimental techniques to allow imaging of ever larger numbers of neurons, try to understand how the brain computes.

I don't see how one could boil down the staggering complexity of the field in the one-bite popsci factoids that leave a layperson with any useful model of the brain. People certainly try though (ahem, Andrew Huberman...).

Here are some facts though: you have some neurons, they do things sometimes. we have some vague ideas about what different bits of brain do in very specific context. we have absolutely no idea how it all fits together, and are decades away from such an understanding. the fruit fly visual cortex, which is several orders of magnitude smaller than mammalian cortex, has been exhaustively studied for 40+ years, and we are only recently starting to build a circuit level model of the computation in the first layers. Here is your fact: brains are the most complex objects in the known universe

andbberger··on Cargo sous terrain – Underground cargo train in Switzerland
neither will ever happen
← PreviousPage 10 of 32Next →