feelix

I'd certainly like to think that's the case, although unfortunately it does not seem to be. If you read the article I posted in my initial comment, for example, it seems like this involves value combined with PR BS.

If you think that's not the case and it truly is a game changer please do share why

How would you consider a drug that cures 5% of cancers a "game changer"?

Do you know what a game changer is? Hint: It means the game changes. In this case it would mean that cancer is at least 50% obviated.

I couldn't find them, sorry

It seems like the PR engine is on overdrive for this drug. And it looks the BBC are spewing it out without giving it any thought at all. Or perhaps the reported was paid?

I'm not saying it has no value, I'm saying that their PR is wildly out of control and it's reflecting on the poor state of journalism more than the efficacy of the drug.

Here is a scathing review of it and the PR engine behind it from the NYT just a month ago, ironically: http://www.nytimes.com/2016/08/09/opinion/cancer-drug-ads-vs...

OPDIVO is another name for nivolumab.

No one should call a cancer treatment drug a 'game changer' unless it literally cures cancer in something like 90% of cases. Otherwise it's spreading false hope and it is actually immoral - if you consider that nearly everybody has been affected by cancer either directly or indirectly.

Additionally the apps outside the app store tend to work a lot better (nearly all the apps I wrote work far better in the website versions, not having the restrictions of the MAS versions), and also the developer gets more of the money, 95% instead of 70%. And then you're a customer of the developer, not of the app store, which is also better.

I said "relatively". That is to say: relative to getting 3rd party app support from everybody.

This was probably before NeXT

Because it's (relatively) easy to develop a good OS, or one better than Android, but it's going to be very difficult to get the 3rd party apps to support your platform before you have enough users. And you wont get the users without 3rd party app support.

I find it odd that you'd criticize Sam Altman in this way then go on to call Steve Jobs a "true visionary". I see Jobs as being more a managerial type that managed to get a few things very right. He wasn't able to dive deep into technical things and grok them as you said. I have read comments on hacker news from people that worked directly under him, talking about how he forbid using object orientation in programming. So programmers had to hide it from them that they were actually using it, which was easy because he could not read the code.

Additionally, Sam Altman is now in a position to achieve greatness and he probably will do so whether or not he has already. Someone in that position would not even have to be particularly exceptional (once again, whether he is or not) to do so. So whether he has or not so far is a moot point. I'm just pleased he has ambition and apparently isn't focused on the lime light. Compare that to how most billionaires use their funds (ie, big yachts etc) and you should just be pleased that anything is happening.

Anyway, my main point is that I guess that an appreciation of "genius" is a highly subjective thing. And the instant someone would mention Jobs as such is the same moment I'd intensely disagree with them. It can probably just be left at that.

On that note I've always wondered if putting an SSD in an oven at the highest temperature just before damaging it could revive it from being written to and erased too many times.

The same mechanism that makes SSD's not retain data (that is, the electrons escaping) and the same thing which makes them not retain the data that you do want to keep would presumably also make the electron buildup that you get that kills the cells abilities to continue their write/erase cycle.

Is it possible to run customized (ie, recompiled, with new C code embedded) TensorFlow instances in it now? without that there's no point in looking.

Thanks for replying.

Yes, a lot. but I don't really have a permanent location to store the server unfortunately.

Using spot pricing you tend to get it far cheaper than that. I was using the previous GPU instances a few months ago which are meant to be $0.65 an hour. But using spot pricing, setting the maximum to about $50 so I wouldn't get shutdown in the middle of training, I seemed to spend an average of about $0.20 an hour.

Now I'm using a P2 instance to see how much faster it runs the same neural net. I set the max price to $0.50 in a spot price bid and got it immediately. I'm not sure how to find out exactly how much it is being charged however.

I tried using google cloud (switching from AWS) a few months ago and it was the worst user experience I've ever had trying to use any software / dev environment. The UI was far worse than windows 98, and horribly confusing. And I got the impression you can only run standard tensorflow libraries, without any modification. I needed to run a modified version of tensorflow so it was a massive waste of time.

I couldn't believe how much better AWS is for running TensorFlow models than Google Cloud.

You can set up the volume a CPU only machine (even on a free instance), and then just launch that volume with a machine with these big expensive GPU's on it.

I did it a few months ago with a TensorFlow model I was training. Configured it for free then all the training time was fully used after configuration.

I assume it'll be the same with these new instances. I have a backlog of stuff I wanted to train but had no GPU to train it on (I travel a lot so couldn't buy a desktop, and was almost biting the bullet to buy one of those monster laptops with dual 1080's, although really I was after at least 12GB of VRAM so this is fantastic news for me)

My god, this is brilliant

Yes, and they're disproportionately more expensive (ie more than twice the price of 2 500GB's) because of the aforementioned issues. It's also the bottleneck in having say 5TB SSD's that are just 20 smaller ones stuck together, because you could cram them into a 3.5 inch enclosure.

You generally can't just stick separate chunks of flash memory together to make one big one, because the controller is the limiting factor. Having a chip being able to manage large amounts of flash memory (having to have its own memory for the allocation tables, wear leveling, and so forth) is where it gets tricky.

>- I am REALLY bad at algorithm problems and experience serious performance anxiety.

For me this is the key sentence here.

Don't worry about feeling like you had to work harder than everybody else during college, that is called imposter syndrome and it's normal.

As for the performance anxiety on technical stuff, I had the same thing. I was even barred from being allowed into programming courses, that I would have to pay a lot of money for, because I failed the aptitude test. I went back in days later and I demanded to retake it after failing and was allowed, and did even worse on the same test.

So I had crippling self doubt. Then given a few years to work on my own stuff, I taught myself C and became the most prolific developer (my work is linked to in other comments I've made on HN if you want to check it) I know churning out loads of awesome software. I even got into kernel hacking in C and stuff. I still don't know if I'm inherently clever or even a good programmer, but I do know I've achieved good results. I think if you're passionate about it you'll find a way to create what is important to you, even if it's very technical.

Obviously the assertion here is that Scandinavia will roll over for the US.

> The extradition to the US seems to be Assange's own narrative

This is the real issue. Whether or not he would likely be extradited if he were to return to Sweden.

Obviously Assange thinks he will otherwise he would have returned long ago.

If you think he would not be, then that should be the focus of your debate.

It's all using Apple endorsed formats. It's just using an awful lot of them. But it's all exactly how it's supposed to be. But because it's using a daemon, a Finder extension, command line interface apps, etc, they all have their own provisioning profiles and they all have to be codesigned in their own ways. Which is what makes it tricky.

> and preventing cross-platform development.

One of the features of Swift is being cross platform. A good 20% of the page that this post links to is talking about their cross platform progress for linux and windows. https://swift.org/about/#swiftorg-and-open-source

That's probably a good approach. Still, when my certificates expire (which makes the provisioning profiles invalid), that alone will take hours or days to fix. So even if it's left in a completely untouched environment certificates expire after a year or two and I cannot recompile the project without a lot of work.

I have a large complex project in XCode. (It's this: http://macdaddy.io/mac-backup-software)

It looks simple but it's actually about 6 different programs all stuck together.

Codesigning breaks on it periodically, and it takes days to fix.

XCode 8 has just broken this worse than any other XCode version in history. It seems to have done so by trying to automate all its requirements. So now it outputs broken bundles when I compile it. This has taken me all of today, and I'll be surprised if I have it working within the next 6 hours of work.

This is a fairly typical experience of using XCode for me. Many times getting codesigning working correctly and getting it working within the mac app store parameters takes more time than writing the code.

In short: Using XCode is harder than programming.

Yes, that's me. Hacker News is the only place I feel I can talk about this kind of thing so it's cathartic. And no, it doesnt work on SSD's. They're black boxes. However I think data recovery on them is possible with hardware. It's the controller that returns 0's after it has been sent the command to TRIM some free space.

This is the data recovery app consumer version that kicked it all off (it has the same engine as the original one): http://macdaddy.io/mac-data-recovery/

You wont find any data recovery software that does a more thorough scan than that, even though it was written in 2004 or so and has been barely modified since. I spent years writing it, in C, so it ended up highly optimized. In forensics mode it scans a drive byte-by-byte, and it assumes each byte is the start of a file of the 100 or so filetypes that it supports, and it maintains that assumption until it is proven to be false as it goes deeper into that file's data. That way it really shouldn't miss anything. It also transparently UUDecodes, unzips, and otherwise decodes data on the fly as to extract file contents under as many conditions as possible. It uses linked lists to maintain its memory of what files are where until it dumps out its buffer as output files, and it uses binary searches and threads to speed up. All in all it's about 1000 lines of code, though I really don't think that anyone would ever be able to read it except for me.

I recently noticed that if you run it over your iTunes Music folder it'll extract and output the media that is contained inside your M4P ("MPEG 4 protected audio") files, and output it as unprotected MP3's. So it effectively strips all of the DRM off of all your media in seconds. Even if you didn't purchase that media (say your friend sent it to you) and couldn't play it before. That was not intentional. You can see that effect in this app: https://itunes.apple.com/app/file-extractor/id1129674765 which is the same codebase, but it is being applied to used space (files) instead of free space.

I wrote some really effective data recovery software a decade ago. Before I knew it governments around the world were buying it up in droves. I then got pushed (not by the governments, but by my partner at the company I had) to write a special forensics edition, that does a byte-by-byte scan of the source disk rather than a block-by-block scan, which would extract embedded files out of a hard disk and basically ensure that nothing would be missed.

That code eventually got a big fancy "forensics" UI wrapped around it with a bunch of other functionality such as logging the process, and it got sold for $1k / copy. It would sell upwards of 80 copies at a time in a single batch. It was very profitable (though I personally did not see much of that profit).

I always felt very uneasy about it and I stopped developing that line of software even though at this point I could get ~50% of the profit if I were to continue it. I had no idea what it was actually being used for all of the time and I had no way to find out. I do know however that some of the time it was used for good, to catch people distributing child pornography and so forth, however I don't know what percentage of the time it was used for that kind of thing, and I'm also aware that that is the justification that is used for a lot of surveillance. Philosophically I believe in counter-survelliance more than surveillance, because I'm pro privacy and pro citizen empowerment rather than the other way around, I think the balance of power has gotten out of whack as this article nicely illustrates.

It does use a hell of a lot of RAM, but not that much usually. For a 1MP image it'll usually use ~8GB. I think there's something else wrong. What sort of machine are you using? & OSX version?

Aside from all the other issues with this, you need a visa to get into China. And you need a ticket out to get in. I imagine this applies to you even if you're pretty high in Apple management.