Hacker News new | past | comments | ask | show | jobs | submit | simonpure's comments login

The foundation of PageRank is counting incoming links to a website. Linking from your website to another website is largely a human decision and PageRank aggregates all these decisions across the Web.

You can either optimize a search engine for a specific vertical or you can try to build features that generalize across the Web no matter the existing verticals or future ones.

While Google is certainly looking at the problem as an algorithmic one, it very much relies on human input to decide what is important and relevant.


There's also PeerReview: Practical Accountability for Distributed Systems [1] published in 2007:

PeerReview ensures that Byzantine faults whose effects are observed by a correct node are eventually detected and irrefutably linked to a faulty node. At the same time, PeerReview ensures that a correct node can always defend itself against false accusations.

[1] https://doi.org/10.1145/1323293.1294279


Author here. Yes, PeerReview is the most important piece of literature for us. We use a lot of similar concepts. Difference is that we're working for the past 18 years on getting this fully working, and deployed. Science is slowly making progress towards an "Internet-of-Trust".

- PeerReview assumes full network knowledge which is unrealistic in open p2p networks.

- PeerReview is built for full protocol replay and replication whereas TrustChain selectively replicates relevant information in the network. This significantly reduces storage and communication overhead, at the cost of fraud cases staying under the radar.

Latest work is our upcoming AAMAS 2021 publication, "Achieving Sybil-proofness in Distributed Work Systems", https://staff.science.uva.nl/u.endriss/aamas-2021/programme/...


From the Bloomberg Green newsletter -

Consider an example. LCCA value for rooftop solar varies hugely across the U.S. If the installation helps displace a low-efficiency gas turbine in California, then LCCA is only $60 per metric ton of CO2 avoided. If it replaces a high-efficiency gas plant in New Jersey, then LCCA is $320 per metric ton instead. Following this logic, each dollar provided in federal tax credits to incentivize rooftop solar—which is currently given out uniformly across the U.S.—goes much further in cutting carbon in California than in New Jersey. In effect, if LCCA were the only metric and cutting greenhouse gases the main goal, then perhaps California should get more federal support for rooftop solar than New Jersey.

...

Still, LCCA provides insight into what can bring the greatest amount of carbon reduction for the same price. It gives us a framework to make a more apples-to-apples comparison of policy or portfolio options to decide which will be the most effective in cutting greenhouses gases, and thus which will have the biggest impact in addressing climate change.


This is also known as the Socialist calculation debate [0].

[0] https://en.m.wikipedia.org/wiki/Socialist_calculation_debate


Google published their guidelines for designing APIs [0]

[0] https://cloud.google.com/apis/design


Matt Levine covered this same strategy back in February wrt WSB [0].

Too bad he's on paternity leave at the moment so we won't get his take this time around.

[0] https://www.bloomberg.com/opinion/articles/2020-02-26/reddit...


I've been missing Matt Levine. He needs to get back already! Best newsletter on the internet, hands down.


From the article:

> When shares keep rising, managing the hedge entails buying more stock.

I don’t understand. If I sell a call option on a single share, I can hedge this by buying a single share and not doing anything until the contract expires, right?


Sure you could, but that's not how the dealers are (usually) operating. The option is much cheaper than the stock, so when selling a call option you're unlikely to have the entire position covered. From further down the article:

"Options are a way to get leverage, too, and can work sort of like an extreme version of the margin loan. If you have $100, you can buy options on $1,000 worth of stock, and a dealer will go out and buy $500 worth of stock to hedge that option."

The dealer doesn't own all of the stock they have sold you the option on. So when the stock price goes up, their exposure goes up, and they have to buy more stock to manage it.

If you're happy to own the entire position, why sell the call option?


The New York Times DealBook newsletter had a link to it in today's issue [0]. It's at the very bottom.

[0] https://www.nytimes.com/2020/08/13/business/dealbook/stock-m...


I was looking into the same issue and learned that modern browsers now also support SVG images for favicons.

So far, this is what I've come up with -

  <link rel="icon" href="data:image/svg+xml,%3csvg%3e %3c/svg%3e">


Interesting solution! It shows at the "no favicon" image on the page tab (the dog-eared page icon) in the latest Chrome/Edge for me, not as a blank/clear icon. (I get that same result when I used GIFs that were invalid).


Thanks for letting me know. I probably have to tweak it a bit to make sure it's valid in all browsers.

While researching, I've found some fun things people have been doing using text and emojis in SVGs [0]

[0] https://lean8086.com/articles/using-an-emoji-as-favicon-with... [1] https://css-tricks.com/emojis-as-favicons/


Another interesting project I've learned about recently is Snowflake [0] using WebRTC datachannels to use the browser as a proxy to help with additional entry nodes. Technical details are at [1].

I have no experience using it for any extended period of time so use at your own risk.

[0] https://snowflake.torproject.org/ [1] https://trac.torproject.org/projects/tor/wiki/doc/Snowflake


There's the Web Transport proposal. It provides client/server streams similar to what you can do with data channels (ordered, unordered, multiple streams) and it's using QUIC -

https://wicg.github.io/web-transport/


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: