Hacker Newsnew | past | comments | ask | show | jobs | submit | more lambda's commentslogin

It's more than just the size.

If you pull in lots of dependencies, people think "great, now there are a whole bunch of dependencies that I'll need to keep updated, may break my program if they update in the future, may not be available (or available in the appropriate version) on another platform if I switch platforms, etc."

Dependencies have a cost that is more than just size. There's the extra overhead in managing them any time you need to do something beyond the first time installation.

Sure, if dependencies are all appropriately semantically versioned, with proper symbol versioning and sonames in libraries and so on, and you have a proper packaging system which keeps all of those dependencies updated properly, the cognitive overhead is not that high. But people screw these things up all the time; making backwards incompatible changes without incrementing major versions or updating sonames, packaging systems that don't allow you to have two incompatible versions installed at once, etc. And the larger the dependency set, the more likely it is that you'll run into once of these problems.


  He works in an industry that does not allow you to email 
  code home or copy it off of company computers.
If you read through some of the subpoena petitions in the docket[1] such as [2] you'll find that the defense was asking for documentation of conversations indicating that it was a normal thing for him to take work home with him, and his supervisors were aware of it. Since most of the evidence is under seal (as some of it contains trade-secret information), it's hard to say if he found the evidence that he needed. But it does look like it he's at least claiming that it wasn't unusual for him to take his work home with him.

For example:

  The defense will demonstrate at trial that Aleynikov
  regularly transferred files offsite and to a local 
  directory on hosts at Goldman’s development network so 
  that he would be able to continue working productively in 
  case of a network outage at Goldman to servers hosting 
  repositories with platform source code. The records 
  requested in Item 21 will demonstrate that shortly before 
  Aleynikov left Goldman, there was an outage that made 
  access to source code unavailable at least for several 
  hours that impacted developers’ productivity. This
  information will aid the defense in demonstrating that 
  Aleynikov did not have the specific intent necessary to
  commit the crimes with which he was been charged.
[1]: https://ia600209.us.archive.org/9/items/gov.uscourts.nysd.35... [2]: https://ia600209.us.archive.org/9/items/gov.uscourts.nysd.35...


Yes, I believe that the license is for the corporation as a whole, so it using it internally does not count as distribution.


If anyone's interested, I found the appeals court ruling finding him not guilty of violating federal law, and leading to his release, on PACER, uploaded to the Internet Archive so you can access it without a paywall via RECAP:

https://ia700209.us.archive.org/9/items/gov.uscourts.nysd.35...

The basic argument hinges on the three counts, related to three laws he was alleged to have broken: the Economic Espionage Act, the National Stolen Property Act, and the Computer Fraud and Abuse act. The third charge was dismissed by the district court because it rested on the fact that he had either accessed systems he was not authorized to or exceeded authorized access. However, he was authorized to access the source code in question, and what he did with it afterwards has no bearing on whether he exceeded his authorization, so it doesn't fall under the CFAA.

The district court did convict him on the first two counts, but the Appeals Court reversed. Their argument is that the National Stolen Property Act doesn't apply because it applies only to actual physical goods, not mere intangible ideas. Had he photocopied the source and walked out with it, or loaded it onto a thumb drive at the office and taken that with him, it would have counted as stealing a physical good, but merely uploading it to a server and downloading it onto a thumb drive later does not count.

The court further argues that he did not violate the Economic Espionage Act because the clause in question he was prosecuted under specifically requires that the "trade secret ... is related to or included in a product that is produced for or placed in interstate or foreign commerce". Since Goldman Sachs' HFT trading system entirely proprietary and internal and not produced for or placed in interstate commerce, that particular law does not apply. Apparently Congress specifically intended this restriction, because earlier drafts of the statute had broader language that merely included "proprietary economic information having a value of not less than $100,000". The fact that Goldman Sachs uses the product for interstate commerce is not compelling, it had to itself be produced for or placed in interstate commerce.

That last part is interesting. It implies that if you run proprietary, internal code that is not sold or intended to be sold in the future, you appear to lose federal criminal trade secret protections. It's interesting that they tried to prosecute him on theft, trade secret infringement, and exceeding authorized access, but not copyright infringement. From my reading even unpublished work is subject to copyright.

Neither the original conviction nor the appeals court opinion ever addressed the copyright issue. In order for him to have stolen something, it would have had to be something of value; so why wasn't he further prosecuted for copyright violation? From the documents I read (not all are available on PACER), the copyright question never even came up.

More documents from the case:

Motion to dismiss the original case in district court: https://ia600209.us.archive.org/9/items/gov.uscourts.nysd.35...

Government's response to the motion: http://www.archive.org/download/gov.uscourts.nysd.358303/gov...

Affadavit of the investigating officer: http://www.archive.org/download/gov.uscourts.nysd.358303/gov...

District court's opinion dismissing the third count but refusing to dismiss the first two: https://ia600209.us.archive.org/9/items/gov.uscourts.nysd.35...

List of files requested in discovery, to demonstrate that what he took was insubstantial and not proprietary: http://www.archive.org/download/gov.uscourts.nysd.358303/gov...

A few of the things he had downloaded were their version of the Erlang platform, which is available under the Erlang Public License, a derivative of the Mozilla Public License. So it would be more fruitful to debate the merits of that license, not the GPL.

All of the currently uploaded items in the docket: https://ia600209.us.archive.org/9/items/gov.uscourts.nysd.35...


I may be understanding the laws wrong, but I believe that he wasn't prosecuted for copyright law because copyright law mainly deals with redistribution, not personal use. Even the DMCA in its worst interpretation only disallows circumventing DRM to make a copy for personal use; making a copy for personal use by itself has never been covered by copyright law. Since he wasn't planning to redistribute the code he copied, he wasn't affected by copyright law.


Is he on trial again after an appeals court already found him not guilty? The article is a little unclear on that point. First it says he's suing Goldman to recoup losses, and then at the end it says:

Aleynikov has pleaded not guilty and is waiting for his trial in New York State Court.

Does "not guilty" refer to the trial in which he's suing them? Or is there another trial even though he's already been found not guilty (double jeopardy)?


Yes, he's on trial now for violating state laws, since the appeals court ruled that the federal laws in question did not apply to his case.

So, he's been found not to be guilty of those particular federal crimes, but there may be state crimes that he could still be found guilty of. They are separate jurisdictions and separate laws, so it doesn't count as double jeopardy.


United States recognizes a legal doctrine of dual sovereignity, so a state can try what the federal government already tried.


...also known as "we carved yet another exception into the clearly worded text of the Constitution"


The Bill of Rights, like most of the Constitution except clearly identified provisions, wasn't originally intended to apply to the states. It was only in the 20th century that the Supreme Court started the process of incorporating provisions of the Bill of Rights to the states via the equal protection clause of the 14th amendment (which was a reconstruction amendment explicitly imposing restrictions on the states).

So the clear text of the Constitution, as the drafters understood it, did not prohibit concurrent federal and state criminal trials for the same crime. Indeed, it imposed no limits on state criminal procedure at all. No due process, no nothing.


IANAL, but if I understand correctly, to claim copyright infringement, they would have had to register the work with the library of congress, and essentially make it publicly viewable - which I assume they didn't want.

Copyright in this kind of case is usable almost exclusively by GPL style licenses, where the code is already available.


Registration is not required.

Registering allows you to sue for more money than if you don't register, but any creative work is automatically protected by copyright law (including full rights to control and prevent reproduction/distribution) automatically as of the moment the work is first fixed into a "tangible form of expression".


> Registration is not required.

That's not entirely correct. From the U.S. copyright statute, 17 USC 411: "... no civil action for infringement of the copyright in any United States work shall be instituted until preregistration or registration of the copyright claim has been made in accordance with this title." [1]

(The term "United States work" is defined in 17 USC 101 [2]. There are certain exceptions, unlikely to be relevant.)

[1] http://www.copyright.gov/title17/92chap4.html#411

[2] http://www.copyright.gov/title17/92chap1.html


So if you find out someone is infringing and want to sue, you begin registration for copyright and file suit shortly after. It doesn't say it had to be registered when the infringement happened.


> So if you find out someone is infringing and want to sue, you begin registration for copyright and file suit shortly after. It doesn't say it had to be registered when the infringement happened.

Technically that's correct -- but there are some possible adverse consequences if you don't register before the infringement begins (or alternatively within a three-month window after first publication). Specifically:

a) you forfeit the option of seeking "statutory damages," and instead you have to prove up your actual economic loss and any additional profits made by the infringer arising from the infringement; and

b) you don't get to seek attorneys' fees from the infringer if you win --- but if the infringer wins, the infringer can still seek attorneys' fees from you.

See http://www.law.cornell.edu/uscode/text/17/412


Registering allows you to sue for more money than if you don't register,

Maximum is $250,000 per infringement if it is registered (and possible jail time for 10 or more infringements). If it is not registered then liabilty is just (simplified version) the retail price of each work infringed.

So, without registering it and without ever pricing it for sale, it could be argued that the damages would be $0.


And even if registering, he only made one copy, and $250,000 is kind of cheap for what such a thing may be worth....


Thanks. But wouldn't they still need to make a copy available e.g. In the court documents? That would definitely be an angle they would want to avoid


No, they are not required to make a full copy available to the court in order to enforce copyright.

The US Copyright Office has a FAQ document which explains what registration does:

http://www.copyright.gov/help/faq/faq-general.html

A more detailed explanation is in this pamphlet:

http://www.copyright.gov/circs/circ1.pdf


It's possible to do a warm reboot into a new kernel without ever powering down and re-initializing RAM.

Also, it may not even be fully rebooting, but instead doing something like killing all processes except init (launchd) which then starts up the installer. I haven't run it, so I don't know exactly what it's doing, but there are several ways to "reboot" without fully rebooting.


It's not a new idea, and it's not a bad idea.

These kinds of things are best handled as an integrated component of the operating system, rather than a hacky third party add on, and are likely more viable now that we have so much more available CPU time, on multiple cores so it doesn't even impact latency, and better compression algorithms.

Also note that Ram Doubler, per the Ambrosia Software description, didn't just do compression, it also re-used free RAM that other applications had allocated to them while they weren't running, depending on the cooperative multitasking nature of the operating system to avoid conflicts. Before yielding, it would either compress or swap out that "borrowed" RAM.

A properly virtual memory system and cooperative multitasking make these techniques not particularly useful, while improvements in technology since have made compressed RAM more useful.


Hmm? It did describe the major reason, and it wasn't younger people killing themselves in accidents or infant mortality:

  Right away, they found our weak spot. "U.S. women have 
  relatively high mortality rates at the younger older ages,"
  they said, which means when women hit their 55th birthdays, 
  for the next almost 20 years, roughly 55 to 75, they will 
  die more often than women in comparable countries. Americans
  get more lung disease, more heart disease, more diabetes. If
  Americans reach 75, they get competitive again, but that 
  early old age is where we lose ground. American men showed 
  pretty much the same weakness at roughly the same times.
They just haven't been able to explain why this discrepancy at this age exists.

Did you not see that part, or did you have some disagreement with it? I have not read the full report that it was based on yet, so I can't comment in detail: http://www.ncbi.nlm.nih.gov/books/NBK62373/


I think one of the key points about food science is that you should always be skeptical of it.

In general, it is very difficult to do controlled, large population studies over a human life span. Instead, lots of studies rely on self-reported surveys, measurements of various indicators that are believed to correlate with poor heath effects over a very short period, and so on.

Mapping from "we know that if you eat X, the Y level in your blood rises over the next 24 hours, and in a separate study we have shown that Y is correlated with chronic disease Z" is actually a pretty poor argument for "X causes Z", and even more of a poor argument for "thus, you should eliminate X from your diet", and telling people to eliminate "X" from their diet doesn't necessarily always lead to simply the desired effect of eliminating "X", but may lead to them replacing it with "W" which is even worse.

Human bodies and human behavior are quite complex. The actual effects of a particular recommendation are not well understood. By making recommendations and policy from science that was not well enough understood, we may have increased the rate of growth of the obesity epidemic over the past 30 years or so. This has happened, in part, by over-emphasizing "low fat" which encouraged people to replace fat with sugars.

So, food policy recommendation are not food science. They may be influenced by science, but they are different. And I generally take them with a very large grain of salt.

Instead, we can find strategies that, at least anecdotally, seem to work better than adhering to the latest fad diet or the government recommendations. Eat real food; plants and animals that have been minimally processed before getting to you. The more processing that has been done, the more the foods have likely been optimized for something other than nutritional value and flavor (shelf life, profits, etc). Don't eat too much; whatever you do, if you want to lose weight, eating less across the board will likely achieve that. Food that is decadent is likely good to have only in small quantities; if it's sweet or greasy, it's probably good to limit to lower quantities. Get exercise daily; walk, or bike somewhere, hit the gym, do an outdoor activity, or something, don't go from bed to car to chair to car to couch to bed.

Following these kinds of recommendations may not seem as official or authoritative as the government or some author selling a new diet book. But anecdotally, they work pretty well, and they aren't likely to be skewed by various parties monetary interests or simply bad interpretation and over-generalization of good scientific data.


It's an early version, and for many uses it's more stable and less intrusive than Reader.

I have avoided Reader for years by using Safari, and later Chrome, which each have their own built in PDF renderers. I'm glad that Firefox has caught up, it's one of the things that has kept me from using Firefox. I read a lot of PDFs, and loading the big, slow, clunky Reader plugin, or downloading the PDF, is a non-starter.


Yep, banned on OS X, kind of frustrating since I need to use Java for work occasionally, and when I click on the link and it redirects after opening in Chrome I need to go back, copy and paste the link into Firefox.


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

Search: