cjf101

The one I've seen (and been impressed with) uses http://smmry.com/ to condense articles to a tldr.

That's pretty dead on to how I read HN. Full articles when I am curious about the subject matter, straight to comments when I am more curious about how others view the material. Sometimes the comment threads will prompt me to go back and read the source.

That is blackly humourous to me. Wow. Thanks for the link.

"The researchers found that certain intensive procedures were performed less often on the high-risk patients in the study during meeting dates than outside meeting dates.

One explanation for these findings, the researchers said, is that physicians who don’t attend the conferences take a more conservative approach for high-risk patients; another is that the physicians who stayed behind were reluctant to perform intensive procedures on another physician’s patients while that doctor was out of town. Survival rates might be higher because, for high-risk patients with cardiovascular disease, the harms of intensive procedures may unexpectedly outweigh the benefits."

This sounds a little like the programming conference effect. Where someone comes home excited about a new way of doing things, and applies it without first proving it out in isolation. Except people die.

While there may be overvaluation of appearance/social ability, it's likely exaggerated in this particular example.

Appearance and social ability in surgical teams are important because surgeons are also leaders and caregivers. These attributes will matter to the members of their team and their patients. Team members who believe in their leadership tend to be more effective. Patients who believe in their care have better recovery rates.

If you did find such an example, it would be worth finding out what they are doing in other areas to compensate for the lack of this attribute.

Those sensors do fail, however.

http://globalnews.ca/news/2117570/alberta-oil-spill-when-fai... (2015)

Notice how many failure reports are not the sensor, but are due to humans checking up on the pipeline. The conditions these pipelines are installed in can be pretty hostile to equipment and monitoring. Going by those numbers, I'd say sensor systems have a ways to go.

The debate that usually goes around is that pipelines fail less often, but their failures tend to spill more oil, in more sensitive areas.

The other dimension, that isn't often talked about, is which transport system has the best chance to improve. Despite their current shortcomings, I think that a system that is dedicated to one job (moving dangerous fluids) has a better safety ceiling than shared systems, like trains and roads.

There's at least two ways to go about learning ML, and I think ideally one should do both. One is to use a high level tool like TensorFlow or Azure ML Studio to experiment with what the tools can do with data. This can get you sufficient competency to use these algorithms in a practical way.

The other is to learn the foundations of those algorithms so you can best understand how to tune, apply and extend them. This is the path to mastery.

I used this recently when I was relearning NN theory (last time I looked at it before 2015 was the late 90's,) and I agree. It's a thorough walkthrough of the math that turned the lightbulbs back on enough for me to write a simple swift MLP NN without consulting other implementations.

Most bakeries use commercial yeast in addition to starter in what they call sourdough. If you haven't already been doing this, you need to ask if there's commercial yeast in their sourdough. You might need to shop around to find a bakery that does it pure.

My partner has a similar condition (not celiac, not gluten sensitive, but 'something in bread' sensitive) and she found that sourdough from locally made starter was way easier on her system than anything from bakeries or the store.

(Source: We've been experimenting with breadmaking at home, so I've learned more than I ever expected to about bread over the past year.)