HNHacker News
TopNewBestAskShowJobs

sarabande

489 karma · joined November 21, 2016

submissionscomments
sarabande··on A note on programmer salaries
A quick search on Stack Overflow careers shows that while there are many Berlin startups that don't pay well compared to U.S. salaries (say, 60k€ range), there are also a lot of multinationals that pay great salaries; I've seen senior dev jobs up to 130k _salary_ here. Competition seems to be heating up so I only expect salaries to rise accordingly.

Here's an example (data scientist, but still counts in my book): https://stackoverflow.com/jobs/148268/senior-data-scientist-...

I know there are also offices of Google, Facebook, AirBnB, Pinterest in Berlin, although I'm not sure which of those have developers.

sarabande··on 100 companies responsible for 71% of greenhouse gas emissions
I agree with your argument that since consumers are the ones who demand the product, they are the reason such companies exist.

However, customers typically have little information on how many greenhouse gases are produced for a given product they buy, and especially not when it counts -- that is, in a store before a purchase -- so they cannot be reasonably expected to vote with their wallet.

One solution might be a combination of:

1) Labeling all products with the emissions that took to create them, so customers can easily vote with their wallet. Perhaps display them next to prices, perhaps with a label like Nutritional Facts on food products, or like a warning when some emissions threshold is exceeded, like smoking risks on cigarette cartons. Customers should not have to watch documentaries or do research on which companies are environmentally friendly and which aren't, but rather have that information given at purchase time.

One or both of:

2a) Subsidizing environmentally-friendly ways of production (perhaps even just for a time) so that these companies can survive against their coal-burning competitors. This way, customers voting with their wallet don't have to pay extreme prices for alternatives just because they think reducing greenhouse gases is good for the world -- with subsidies or tariffs, we could make alternatives somewhat price-competitive.

2b) Taxing goods made with too many greenhouse gases (i.e. the reverse of 2a).

sarabande··on Ask HN: Who are the most controversial (yet still popular) HNers?
One way to calculate this would be to use this controversy formula (https://math.stackexchange.com/a/318510/10887) on all comments of a user whose total karma exceeds a certain threshold, say 1000, and then add up those controversy scores.

The problems is that the HackerNews API (https://github.com/HackerNews/API) only provides a total score for a comment, instead of up and down votes, so you'd have to modify your controversy formula to use an appropriate measure of dispersion instead (https://en.wikipedia.org/wiki/Statistical_dispersion).

I'm not a statistician myself, so I can't help further.

sarabande··on If companies interviewed tech recruiters the way they interview programmers
To give credit where it is due, this especially funny quote is nearly verbatim copied from the original article (https://medium.freecodecamp.com/welcome-to-the-software-inte...) from which "inspiration" for this one was taken:

    It means we do Agile. Everything is in a flat structure,
    except when it comes to salary and responsibilities. 
    We have a Translator Master, that coordinates all 
    the other Translators in Sprints."
sarabande··on Can Neural Networks Crack Sudoku?
I find it interesting that when the NN fails to complete the whole puzzle, it seems to fail spectacularly (github.com/Kyubyong/sudoku#results) -- that is, there aren't a lot of good-partial attempts (90% or above). Does anyone know why this might be the case? Does the first wrong placement of a number in a gap essentially ruin the rest of the guesses?
sarabande··on People Don’t Buy Products, They Buy Better Versions of Themselves (2016)
Are there any books you would recommend for marketing techniques? The frameworks you mentioned piqued my interest.
sarabande··on Deep Text Correcter
You don't need parallel corpora -- the OP was generating the incorrect versions of training data through random perturbation (dropping articles, etc.)
← PreviousPage 6 of 6