Hacker News: huffpopo

New comment by huffpopo in "Like Peter Thiel, Tech Workers Feel Alienated by Silicon Valley ‘Echo Chamber’"

huffpopo — Sun, 18 Feb 2018 19:36:34 +0000

I left SF before the election, I could tell early on that Trump was going to win and figured free speech clampdowns in SV would follow and I knew my work was going to be affected. I worked in search and I remember the scandal within the industry when Yahoo accepted money from the Chinese government to censor pages on the 2008 tainted milk scandal. They charged $10K per link. At the time Google was seen as a beacon of free speech and 'do no evil' which was a nice contrast to Yahoo which is apparently OK with killing babies for a small buck.

Now Google, Twitter, Facebook etc. are all finding excuses to allow various meddling of the search results. There is too much money in it, it is inevitable.

Free speech is important to me because without it we quite literally end up killing babies for very small amounts of money.

New comment by huffpopo in "What May Be U.S.’s First Drone-Linked Aircraft Crash Is Being Investigated"

huffpopo — Sun, 18 Feb 2018 08:38:43 +0000

Ah, I consider them human with human faults and incentives. One really sad aspect is that a major killer of helicopter pilots is cutting corners on preventative maintenance in order to save money. I feel like we need to be honest on what people are like.

New comment by huffpopo in "What May Be U.S.’s First Drone-Linked Aircraft Crash Is Being Investigated"

huffpopo — Sun, 18 Feb 2018 07:57:04 +0000

I'm curios on the 10+ downvotes this received since posting; would someone care to give me a reason. Thx

New comment by huffpopo in "What May Be U.S.’s First Drone-Linked Aircraft Crash Is Being Investigated"

huffpopo — Sun, 18 Feb 2018 07:55:55 +0000

Drones have harder parts and I would expect hardness to play a big role in damage done. But it doesn't even need to be hard, raindrops can also do a fair amount of expensive damage.

New comment by huffpopo in "What May Be U.S.’s First Drone-Linked Aircraft Crash Is Being Investigated"

huffpopo — Sat, 17 Feb 2018 20:11:49 +0000

Helicopters are slow, and they probably didn't have the rotor speed needed to yank the collective.

Hitting a 2kg drone would result in expensive repairs, e.g. $30K for replacement rotor blades and you still might crash.

I probably would have taken the hit but I'm much better off than the average R22 pilot who often operate or razor thin margins.

That said; my bet is that there was no drone. My guess is the student made a small mistake, the teacher took over and made a bigger mistake, and both want to keep their jobs/future carers. Drones make an easy target to lie about and are hated anyway. Not to say there are no irresponsible drone pilots, just currently I think it's more likely heli-pilot error at this stage. My hope is the GPS tracking in the phantom well let us know for sure.

New comment by huffpopo in "[dead]"

huffpopo — Fri, 26 Jan 2018 19:02:43 +0000

No apology needed, it was a subtle joke. It looks we both got flagged for it.

New comment by huffpopo in "[dead]"

huffpopo — Fri, 26 Jan 2018 16:34:11 +0000

I'm not really letting it slide, I was extending the premise by pretending to be a sympathetic genocidal nazi, which I hope is still absurd enough to be presumed sarcasm. I'm aware that the joke may not age well so I was happy to get it in while I still can.

The CoC complaint is amazing and has me really worried about Rust and I wonder where the adults are. Usually Meta Language people are meritocracy driven so I'm surprised to see a feminist infiltration. GoLang I could understand...

https://www.reddit.com/r/node/comments/6whs2e/multiple_coc_v...

New comment by huffpopo in "[dead]"

huffpopo — Fri, 26 Jan 2018 16:13:32 +0000

I'm willing to let her insitements for genocide slide, it happens to the best of us, but what really worries me is her plan to "improve Cargo by incorporating best practices from npm".

New comment by huffpopo in "[dead]"

huffpopo — Fri, 26 Jan 2018 15:56:55 +0000

It's common to back channel that information, I have a hard time believing that there was any chance his name would not have come out.

He was also doxed in the regular sense by left wing trolls (which do exist). In one example he was stalked, harassed, and had his photo taken and released from a rather private event.

New comment by huffpopo in "[dead]"

huffpopo — Fri, 26 Jan 2018 15:25:24 +0000

James Damore didn't leak his memo to the public, he was very much doxed first. It sounds like some people can't take what they dish out.

New comment by huffpopo in "One Bitcoin miner is buying 20,000 16nm wafers from TSMC per month"

huffpopo — Mon, 22 Jan 2018 22:36:11 +0000

If you mean Samuel Brannan, then you make money by diverting church funds to pay violent gangs to establish and maintain a shovel monopoly. Then you lose it all to a real estate crash and a divorce.

So the real lesson is be immoral, break the law, diversify and don't get married.

It's not like he is the only one who thought of selling shovels with huge markups.

New comment by huffpopo in "Simone Giertz makes a living creating shitty robots (2016) [video]"

huffpopo — Mon, 15 Jan 2018 09:35:34 +0000

The whole shtick is about lowering expectations.

New comment by huffpopo in "James Damore has filed a class action lawsuit against Google"

huffpopo — Mon, 08 Jan 2018 20:26:54 +0000

Even if James Damore does not win Google has a number of practices that could be considered embarrassing if they were to be made public. I know a number of Googlers who confide to me privately about their experiences. Traditionally they would not have an outlet to express their grievances because no-one really cares. I'm hoping this lawsuit will give them and others a chance leak some of these practices. I'm interested in what people outside of SV tech bubble will think about it.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 23:24:14 +0000

I was talking about backtesting of the models to see if they would have predicted the past correctly. I.e. You run the model from 1970 to today and measure how close the model prediction is with the actual measurements. If it's too far off then obviously the model is wrong and changes are made to try to improve it.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 19:11:31 +0000

Yes; but it seems your answer to my question is that I must also have several decades of experience.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 19:01:43 +0000

The group leaders were pushed things too far for me to accept as legitimate but not far enough to produce the results needed. We were aware of others that pushed things even further and they got to continue working. This adds a selection criteria bias to those still doing the research which has me trusting them even less. And this was before ClimateGate which made the shenanigans done by other groups public.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 18:52:37 +0000

I make no claim that climate change isn't happening. Being between the modeler and the results meant it was my job to explain the result to the modeler. So criticizing climate change models within the scientific community was my job for over a year and a half.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 18:20:14 +0000

I think you're being too harsh on my former colleagues, they were not evil or even particularly bad people. Hyperparameter tuning is necessary and future pollution is often accidental. The problem is when the tuning is goal seeked and the future pollution is not accidental and these problems are endemic to the industry and science in general. CRU East Anglia of Climategate fame went much further and actually published their results and I'm pretty sure none of them are in jail.

My colleagues left the industry which makes me less optimistic about those who stayed.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 18:12:20 +0000

~2006-2007, and it was a continuous stream of models. I was between the modellers and the cluster so I would rewrite the models to work on cluster.

New comment by huffpopo in "Extreme event attribution is an expanding subfield of climate science"

huffpopo — Tue, 02 Jan 2018 18:00:31 +0000

Like a #metoo movement for scientists. AFAIK Hyper-parameter tuning is fairly normal and necessary given how easy it is for the models to become unstable on their own and future polluting is often accidental. It can come down to judgement calls and difference of opinions as to what degree of intervention is acceptable which leaves a lot of wiggle room If you removed all future pollution and and left hyperparameters at the initial best guess you often end up with crazy results in backtesting. E.g. if our past predictions were correct then we should be living on Venus right now which obviously we're not so we must need different hyperparameters. My former colleagues were trying their best and even they knew their adjustments were going beyond reasonable. They never signed off on the final results so the project was politically useless and subsequently killed. They've all since left the climate change industry disillusioned as well.