Recent submissions

Comment · 9 years ago
I agree. Who runs arbitrary code in a JVM these days? We have real hardware based virtual machines now, making this irrelevant.

Back in the day Java JVM was the best way to run untrusted code close to native speed. That hasn't been the case for many years now...

Comment · 9 years ago
Yeah same here, can somebody find us a pirate bay link?
Comment · 9 years ago
For the love of God it's not the traffic its lack of public transport. Once you hit a certain density CARS_DONT_WORK . Having lived in NYC, visiting LA was a joke.

The first thing I did after my first cab ride from the airport was look up public transport since I noticed traffic was worse than NYC. There wasn't any

Comment · 9 years ago
What we seem to be running into, is that any strategy based on trusting some sites and not others breaks when very large groups of sites have opposite opinions. Depending on your starting set you will end up with wildly diverging trust scores
Comment · 9 years ago
Seen the pictures of guys in boats floating out in the kamiokande? I always thought it was the coolest looking experiment.

I was really sad when all the bulbs broke years back :'(

Comment · 9 years ago
Are they? This news saddens me greatly.
Comment · 9 years ago
Fake news is just a new version of spam sites. They mostly exist to make money or enforce an agenda.

People figured out that Google's algorithms could tell when things were generally factually false, unless those things were recent news.

This leads me to believe their algorithms largely relied on news as a source of truth. They're going to have to do something about that

Comment · 9 years ago
Does that change anything I said? All this marketing doesn't seem to lead to anything tangible unlike similar releases from other big companies. It's just fluff.

I'm aware the IBM does consulting and that's possibly the most boring software realm there is.

Comment · 9 years ago
https://www.google.com/amp/s/nakedsecurity.sophos.com/2015/0...

Not the FBI per se, but it shows that someone is clearly attempting to compromise TOR users.

Also there's been whispers about it forever. Much like the "black rooms" at datacentres before all the NSA leaks.

The FBI has a long history of tracking down and compromising CC theft and CP rings, along with silk road and the hoards of clones. Most of these sites are primarily or only accessible over TOR.

Running compromised TOR nodes would be an extremely cheap way to monitor a large portion of illicit Internet traffic. The frequent busts are usually attributed to other reasons to shift attention away from TOR, but this is classic parallel construction.

The feds will nearly always get you on secondary evidence when the primary means is too sensitive... See stingrays. The sheer number of TOR based site busts however is telling.

Anyone relying on TOR for security is a fool. It's more heavily monitored than the regular net.

Comment · 9 years ago
If anyone had one it would be Google :)
Comment · 9 years ago
Can somebody tell me something cool IBM actually does for real besides run softlayer?

I'm so tired of endless waves of marketing bullshit with no substance. At least when most companies announce something it's a tangible product I can actually expect to use at a non ambiguous point in the future

Comment · 9 years ago
You can see from this that google considers text earlier in the page more important.

This page is odd because the false answer is presented first, so google fails to see it. It could tell that the form of the sentence was an answer to the question, just not that it wasn't the right one

Comment · 9 years ago
I'm no wizard, but in writing blog articles I've found ways to fool Google into believing me.

The problem with their algorithms.... is that all that statistics in the world can't help you when you're listening to a guy telling the truth vs an equally good liar.

I can tell you what Google doesn't have, a strong AI. It thinks it knows "facts" but these are merely patterns, and these can be gamed.

Because Google still lacks a strong, truly thinking AI, they rely extremely heavily on statistical models to rate content.

So how do you cheat google search?

Google's systems attempt to figure out the topic of your writing, the style, and quality. Is it scientific? An opinion piece? News? Is it a technical topic? A playful one? Fiction or nonfiction ?

The quality classifiers are much easier to game than topic and style analysis. They determine things like reading level of text but also things like the number of rare nouns, number of technical words, number of typos. Readability as far as font and formatting. Trustworthy signal of your domain and possibly the company and people they determine to be linked with it.

I also have a feeling google uses sneakier signals as well. These include your DNS registrar, phone number, email, and address listed on the site. Who you host with and what technologies you're using. Your mail servers and how trustworthy they are. Geo location, and visitor traffic info as soon as you put analytics on the site (or use amp)

Basically when Google says they have tons of signals, they do. They have a dataset that amounts to every site on the internet for the past 15 years, and they regularly run automated and manual "theory provers" much like quants do with historical stock market data. They find new signals constantly, and run tests to see if their new algorithms are better.

You know how sometimes google randomly takes a bit longer to load search results? My tin foily theory? They'll occasionally guinea pig you on prototype search results to see if they're better. I noticed their response time getting really bad a couple months before the public rollout of new AI powered search for example.

So gaming google? Do exactly everything that a large, legitimate, no-nonsense company would do. From where you host, what you host with, to who you link to. Bonus points if you have significant real looking mail and other traffic from your domain. Extra bonus points if you actually sell something real as cover and do it for at least a few years.

Once you've done enough convince google you're a big important thing IRL...write a ton of really subversive bullshit. Make it sound a real as possible, hell make 90% of it real, just with a single unverifiable fact. Keep pumping this shit out and make sure your garbage is never fake enough to get called out on. Or just make the fake part so hard to verify that nobody will waste the time, kinda like half the science world does when publishing papers.

Comment · 9 years ago
You should remember that the early cloud unicorns are getting super low rates for their name clout as well. Their discounts are passed as extra costs to everyone else.

So while it may be worthwhile if you're huge enough to get a big discount, that doesn't mean it applies to small or medium size companies

Comment · 9 years ago
I believe the truth is somewhere in the middle. The hype will eventually die down and we'll see some kind of equilibrium.

I've been experimenting with "cloud over" where I only use cloud servers for spare capacity. With docker it's getting easier. My baseline CoLo has the same performance as a super high end AWS instance and costs a third as much after my fixed investment.

I still have the cloud if my box goes down or takes a big traffic hit, but I'm not paying for a bunch of cloud servers all the time either.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton