ewheeler

Joined 19 years ago

Karma

165

Submissions

Member since

Mar 29

http://github.com/ewheeler http://scr.im/evmw

View on Hacker News

Recent submissions

Comment · 1 month ago

here's an explanation of their 'computer vision approach' to local weather forecasts: https://jackadam.github.io/2011/how-dark-sky-works/

"We create a 3-channel image where the red channel represents velocity in the x-direction, the blue channel represents velocity in the y-direction, and the green channel represents change in storm intensity" and then use computer vision techniques to predict what happens next

post was written after a discussion here on hn: https://news.ycombinator.com/item?id=3187326

Comment · 4 months ago

MIT's D-Lab https://d-lab.mit.edu/research does a lot of similar work on fuels/cooking, evaporative cooling, and design with locally available materials/techniques

Comment · 9 months ago

I'd suggest commercial aircraft as an even better analogy than cars.

Most of the ongoing costs you mention for cars still apply--but there are also the occasional (possibly dramatic) changes to the interior 'cabin product' like new seats and entertainment systems, new fabrics/branding, new business class seats/pods, changes in seat layouts, etc in order to remain competitive in their market segment. Cars rarely have such significant refreshes, but software products often have analogous design and UX overhauls that are also intended to try to keep the software competitive in its market segment. And again airlines don't need to engage the specific airframe manufacturer like Boeing or Airbus for these, but they do need some semblance of a tech team that have certain domain expertise in aircraft engineering constraints.

Airframes also have major overhauls called MROs (Maintenance, Repair, and Overhaul) about every 6-10 years, which again does not require the original manufacturer but does require significant engineering expertise. To me this is akin to certain ongoing software maintenance activities like updating a codebase to use newer library versions, major database version updates, API or SDK version compatibility, etc.

Comment · 1 year ago

Thanks for posting! Wild that that 20 years ago Clay put a name to these niche tools that are now trivial to make with vibes and an API key.

Also fun to see a mention of Dennis Crowley's proto-foursquare group project from grad school

Comment · 1 year ago

Yes, very dependent on the market. In some East African markets that don't have much history of recycling, Kubik [0] has a neat model of employing folks to recover plastic waste from landfills which are then used to creating building materials

0. https://www.buildkubik.com/

Comment · 1 year ago

While there are some similarities, yes, our brains definitely don't have anything akin to backpropagation, which is the critical mechanism for how current AI models learn.

Hinton has some research on a forward-forward learning paradigm [1], which might be closer to how our brains learn (but the artificial implementations are not great yet). He also posits that maybe the purpose of humans' dreams are generating negative data for such a contrastive forward-forward learning mechanism.

[1] https://arxiv.org/abs/2212.13345

Comment · 4 years ago

Fun fact: Hans Albert Einstein, son of Albert Einstein, helped with the design of the Old River Control Structure.

According to Wunderground's three-part series about the ORCS[1], there is some uncertainty as to whether it will be able to keep the Mississippi River on its current route without failure.

[1] https://www.wunderground.com/cat6/Americas-Achilles-Heel-Mis...

Comment · 5 years ago

Maybe? The Scaling Hypothesis[1] suggests that greater capabilities of intelligence may emerge from scaling up 'scalable architectures' to large sizes. GPT-3 exhibits 'meta-learning' capabilities that GPT-2 did not (like learning how to sum numbers)--probably just because its a 100x larger version of GPT-2.

[1] https://www.gwern.net/Scaling-hypothesis

Comment · 6 years ago

"Malachite Capital Management LLC, a hedge fund specializing in trading volatility, said that it would shut down immediately." according to https://www.wsj.com/livecoverage/coronavirus/card/is2y79bNxh...

Comment · 6 years ago

Correct, here is their Kafka-based stream processing package that presumably ties everything together: https://faust.readthedocs.io/en/latest/

Comment · 7 years ago

Uganda is home to a large regional hub for UN logistics, a state-of-the-art CDC infectious disease laboratory, and similar outposts--presumably due to Uganda's relative stability over the past few decades and strategic proximity to several less-stable neighbors.

Comment · 8 years ago

For those looking for a modern take on improving on the CSV format, I'd recommend Frictionless Data's Datapackage specification[1] which basically consists of a json file of metadata that accompanies a CSV file that describes column types, versions, sources, and how to validate correctness of the CSV's data. This allows for quite a lot of tooling and workflow improvements to CSV files without mucking with the CSV itself

Another hack to improve CSV workflows is OCHA's HXL[2] that is used by humanitarian organizations. Basically adding a row of hashtags in addition to column names, which is surprisingly useful considering the ease of adding these to a file.

[1] https://frictionlessdata.io/docs/tabular-data-package/ [2] http://hxlstandard.org/

Comment · 8 years ago

I'd recommend reading more about the 'information fiduciary' approach[1] promoted by Prof Jonathan Zittrain (and others) which I find to be among the most realistic steps in a better direction, as it offers a 'grand bargain' in the spirit of DCMA's 'safe harbor' for companies that choose to commit to putting consumers and their rights above maximizing profit.

[1] https://hbr.org/2018/09/how-to-exercise-the-power-you-didnt-...

Comment · 9 years ago

Looks like a subset of the open-source https://mermaidjs.github.io/