This would be a better "ping pong balls on a plane" interview question for comp sci candidates.

[1] https://www.python.org/dev/peps/pep-8016/

[2] https://www.python.org/dev/peps/pep-8100/#results

If the business works, I'm thinking the transition from the fund swat team to "real" employees might be difficult as well, probably at a time when you need to be firing on all cylinders.

To play devil's advocate - somehow I drove myself to work today, and the hardware I'm running is just two moderate resolution limited field of view cameras. Not an expert, but from first principles it should be possible to pilot a self-driving car with cameras only, given enough processing power and a smart enough agent. Maybe those last two aren't there in 2018 though.

The direct use of compressed air is interesting. Taking it to the extreme, you could have compressed air as your primary utility piped through your home, and have appliances that run off compressed air. Refrigeration cycle is compression based anyway for example. You would still need some electricity for control circuits for example, but you might be able to keep the heavy lifting air-based.

Also, unlike say internet bandwidth, there is actually a logistical limit to how much alcohol you could consume in a week, and it doesn't seem like it could be much more than 1 order of magnitude over the average casual consumer (3-4 drinks vs. 30-40 per week).

All across healthcare, providers, scientists, administrators are faced with millions of tiny choices every day: should we spend more money to improve outcomes. Should we add that extra feature to this device - costs more but improves care a little. Should we use this more expensive material - costs more but is probably safer. Should we do additional testing - costs more but we'll be more sure it's safe. The calculus of incremental value vs. cost is subconsciously seen as inhumane across the industry -- it's seen as starting down that slippery slope that ends with a "dollars per life" number which feels wrong to everyone. Nobody wants to be the person who traded someone's health for a buck.

I'm not necessarily condemning it, I want the best possible care for my children, but I do think that (like a lot of big socioeconomic problems) this comes back to incentives.

Comments URL: https://news.ycombinator.com/item?id=16150976

Points: 2

# Comments: 0

[1] https://www.movidius.com/solutions/vision-processing-unit

Doing image classification, object localization, and homography (given an input image, which of my known template images is matches it and in what orientation).

An alternative between keeping all data in-memory and going full relational database that's worked well for me is Python + H5PY. Write a simple class where your data getters / setters interact directly with the H5 file and don't keep it in memory.

Super fast (data handling and setup), easy on memory, and doesn't lock you into the schema of a relational database.

I'm always seeing places to apply manufacturing process control statistics like control charting [0] and process capability index [1] 'in the wild'.

[0] https://en.m.wikipedia.org/wiki/Control_chart [1] https://en.m.wikipedia.org/wiki/Process_capability_index

The dial-pad may be too difficult to reach on our big screens now, but the momentum required to change it will never exist because it simply doesn't get enough use to be a pain point anymore.

Excellent write up and use of graphics though. The Bell labs part was really interesting.