But here's the interesting thing.

Who owns what? According to the hard core of bitcoin fans, it is based on one principle - that the blockchain is the final word on that issue, at least as it concerns bitcoins themselves. If you have the key, the coins are yours.

If you say "those bitcoins aren't yours, they're stolen!" you're implicitly accepting another standard of property - of who owns what - as higher than the blockchain. Then you concede to the messy passions of society to determine who owns what, rather than the mathematical certainties of the block chain.

Bitcoin fans are a bit two-minded on this. On one hand, they demand that money in anonymous accounts belongs to whoever controls the keys, and it's none of your business how they got there. On the other, some do call for blacklisting coins, e.g. the coins FBI seized when Silk Road went under. The technology actually makes that possible, unlike with cash or even conventional digital payments.

I'm all for convicting Karpeles for fraud (and in fact this article is old news to me - despite lots of anonymous accounts trying to pooh-pooh it, investigations of the block chain made a convincing case for fraud in the weeks after MtGox's fall), but I'm also for recognizing the limits of the blockchain, and I'd like BTC fans to realize it can't be a substitute for government, or even government-issued currency.

This is damage control. What the docs show, is that OTR encryption is safe. You can use OTR encryption with Jabber, but Jabber isn't encrypted by default.

What they presumably hope is that people will use Jabber and think they're safe. And Slashdot editors are being the useful idiots as usual.

Neural net-based Go programs have been tried countless times since neural nets were invented, and losing to GNU Go 20 out of 200 games is very very far from state of the art. I don't know which version of Fuego they used, but if it was rated 4-5k it must have been an old and weak one. It's currently rated 2d on KGS.

Worth noting that KGS is stingy on ratings, and especially hard on bots since matches are self-selected. If there's an exploitable weakness in a bot, people will ruthlessly mine it for rating.

The real revolution is the technique that CrazyStone pioneered: Monte carlo tree search. Even basic MCTS programs can beat GNU Go 90% of the time. CrazyStone was on 7d on KGS for a while, although in september it dropped to 6d (presumably because very strong players found a way to exploit its weaknesses - since programs play consistently, rating shouldn't vary too much otherwise).

Monte Carlo is the way forward. I'm convinced that the basic approach of it (randomized search, weighted with a tree towards positions more likely to be fruitful) is going to stay. Improvements are going to be about which moves to weigh the exploration towards in the playouts as opposed to the tree part.

Just shows how crazy it is to rely on Wikipedia these days.

By the way, anyone else missing comments they recall making in this thread?

So in other words, IT'S A CONSPIRACY!

The person Gamergate started with is an admitted pathological liar.

The rest can be put down to the journalist class' tribalism, and the need to distance oneself from low-status people close to you in social space.

If it was a movie critical of North Korea, that would send a message. But it seems it's just a none-too-deep Hollywood comedy.

What they might do instead, is sponsor a documentary highlighting the dictators' long history of "air nut" behavior and the horrors of their prison state.

The risk associated with driving isn't just for your personally, but also for the life and health of others.

In PKDs story, Deckard is human, end of story. A main point of the book is that the androids really are bad: Even though they can be vulnerable and afraid (Pris Stratton) make great art (Luba Luft), or even fall in love with each other (Roy and Irmgard Baty), they ultimately all are true psychopaths, without a hint of compassion or concern for other people except for their own benefit.

In the book, the idea of an android having false memories implanted and believing itself to be human is a ruse: The Rosen Association claims it about Rachael, but actually she knows perfectly well that she's an android. At another point, the androids go to ridiculous lengths to fool Phil Resch, a somewhat cynical bounty hunter, that he is an android. But they fail.

Of course what a Hollywood director does in his film is his own business. But I'd like to see a dramatization that was more faithful, and went into the moral and religious aspects of the story.

I can't believe it either, but it would impress me much.

Well, then that takes away the main reason to play Nethack - modest as it was in the first place.
I held out hope for Nethack for a long time, but it hasn't been developed in over ten years. The indie/roguelike scene has moved on, and improved immensely.

There are better options today, on just about any criteria.

Oubliette was the inspiration for Liberal Crime Squad, Tarn Adams' (of Dwarf Fortress fame) satirical game about political extremism.

I never said it was the defining feature. I'd say it is the most important, along with randomization.

Have you noticed there are very few games being made nowadays that are three lives, game over, start from the beginning?

Let's call a game that is real time, has permadeath but not random levels "arcade hard". ("Nintendo hard" is a misnomer, Nintendo had far more non-arcade hard games than the competitors, and were among the first to challenge arcade assumptions of what a game should look like)

I propose that if you hate roguelikes, you probably also hate arcade hard games, and vice versa. I guess you have some people fond of twitch reactions and memorized timings who like hard but don't like randomness - in the speedrun crowd, maybe.

DCSS is what replaced Nethack for me. DCSS is every inch as hard, but fairer than Nethack: spoilers will do you little good, boring/cheesy tactics (e.g. grinding) won't help, and you won't die from misclick-type errors as often. Best of all, it has variation in the endgame, since you need three runes to win but there are fifteen in existence. Once you're able to win the game at all, you can start pushing your luck in the extended endgame by getting more runes before doing Zot.

I'm a bit skeptical of the changes made in recent years, though. A lot of the coherence of Linley's design has been eroded along with the improvements.