Toilets on the Shinkansen are more hyginic than my kitchen in Boston ?

Must memorize 314 more kanji by July 7th to take JLPT N3...

@ekaitz_zarraga @thomas Before writing a new distributed search engine that doesn't work, I'd recommend addressing the question of where the ranking signal for results will come from.

1. You can't do PageRank without crawling the entire web (or at least a significant portion of it)

2. You need a lot of users to improve ranking based on previous clicks for the similar queries

Without either 1 or 2, results won't be good enough for daily usage.

@synaesthetica @ekaitz_zarraga @jartigag Satoshi believed that everyone would use their idle capacity for mining Bitcoin, and didn't see ASICs coming. While some people have more CPUs than others, the idea was that it would distribute rewards more evenly among users. Completely fair distribution of mining rewards is still an open research problem...

@ekaitz_zarraga @jartigag @yabirgb Touring completeness is indeed a problem for those who write and use those contracts, but not for other users of the same blockchain using non-buggy contracts.

Besides, the problem of checking contracts for correctness is more constrained than debugging general-purpose computer software: contracts can only operate on inputs and outputs on the blockchain and are by necessity very short programs (typically a few hundreds instructions).

@jartigag @ekaitz_zarraga

This is a decent explanation of how ZCash improves privacy over Bitcoin: https://z.cash/technology/zksnarks/

TL;DR: a blockchain is a public ledger, and thus all transactions are visible with amounts, source and destination. zk-SNARK is a method for encrypting transactions, but in a way that all nodes can still verify that the transaction is valid (not spending more money than you have, not double-spendoing the same money, etc).

I got hungry around midnight, so I roasted three blocks of mochi in the fish oven and dipped them in soy sauce. Simple and delicious.

https://www.youtube.com/watch?v=C5C8GQnI7C0

@ekaitz_zarraga Last year I took a 2-day crash course in machine learning, and it was my first time using Jupyter Notebook.

We could edit and run the inline Python snippets directly within the browser, and the output would show just below in the form of tables or graphs. So easy!

But what totally blew me away is discovering that every time we pressed the little play button, heavy computation was being sharded and dispatched to GPUs in a remote datacenter.

Hey @Gargron, would it be possible to generate html fragment anchors for individual comments in a thread?

This way, your friends who don't have a Mastodon account wouldn't have to scroll down in a long thread to find the highlighted comment.