One of the most fascinating posts on HN in a long time with a headline that is almost completely non sequitur. "Erlang scores a marketing coup"? Seriously?
This is a story about industrial sabotage at Goldman Sachs!
Industrial sabotage involving the theft of software written in Erlang which keeps Goldman Sachs quite literally one of the most powerful financial groups in the world.
There are multiple neato headlines that could be written for this, I wouldn't have chosen this one, but it's not a non sequiter at all.
That is very cool. But i'm not surprised. High reliability, distributed computing and maintainability is key in financial sector. I just can't wait for the code to be posted on some torrent site I'd love to see it.
My own personal trading system is written in ADA with spark for reliability and determinism. So I'd love to see what drove them to decide on the technologies they've used.
EDIT: The absolute best part is at the end of the affidavit. The guy tries to defend himself by saying he was just moving some open source code off the system. After tarballing it
up with their proprietary trading system, and encrypting it.
Who wouldn't believe him?
On another note, if anyone wants to hack on a (soon to be) open source trading platform, get in touch.
How could a trade be executed in microseconds considering that light travels 300m in 1 microsecond? It would seem that low latency connections to NASDAQ trading systems would high priced real estate. Is this something that NASDAQ sells?
Yes, the nyse, nasdaq and such rent floor space for systems.
You get a high speed(low-latency), unencrypted (low latency), unpassword protected (low-latency), low-latency connection
with very low-latency characteristics directly into trading engines.
They go to great lengths to ensure low-latency performance.
Currently there'll be tons of hifi hedge funds upgrading
to nahalem processors and measuring how much faster each individual instruction is. Cutting fiber optic wire to the exact length to lower latencies, then measuring to make sure
they don't have any retransmits due to bounceback, etc.
If you're able to reliably make $0.01 on each trade you'll
be more than willing to spend hundreds of millions on hardware. Microseconds matter, because they add up to milliseconds.
I wasn't involved in the real estate side of things so I don't know how we got the space, but I did work for a company that had machines right on the trading floor of almost every exchange in the world. So, I think the answer to your question is probably, yes.
Attention managers! What is the black magic stealth technology behind feared Goldman Sachs and its whispered microseconds-fast automated trading systems?
The author makes wild speculation about "rogue code" but from the technology in use it doesn't sound as much some kind of super-high-tech world-dominating intelligent decision algorithm so much as just a specialised, high volume, distributed message queue.
I guess other firms might be interested in how GS approached the problem, but it's a quite a leap to state that the matter "could (and likely should) become a matter of National Security".
Wow, that's some wild speculation at the end. "Shouldn't all market participants be aware that there is some rogue code in cyberspace that can be abused by the highest bidder, who very likely will not be interested in proving the efficient market hypothesis?"
Doesn't the new owner of this "rogue code" still need money to make stock transactions? Was GS's interest in running the code proving economic theory, or was it making a buck? If you whistle the code into a pay phone, does it launch the nukes?
Was GS's interest in running the code proving economic theory
Ofcourse not. They (or at least the author of the article) are just crying us a river
because someone stole their cash cow. I can only say: If a large part of their profit was truly based on automated trading by that software then perhaps they should have protected it better in first place? Or is homework now optional when you're in a position to cry for government bailouts?
Anyways, if there's really a software to print money (which this article seems to suggest) then could someone please send me a copy? I'm very curious about this efficient-market theory and would really like to help to prove it...
> For what it’s worth, Aleynikov told the cops that he thought he was just accessing “open source” files on which he had worked, and only later realised he had accessed more data than he realised. He’s also denying that he distributed any of Goldman’s proprietary code.
This is a story about industrial sabotage at Goldman Sachs!