<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:media="http://search.yahoo.com/mrss/"
	>
<channel>
	<title>Comments on: When large-scale complex IT systems break</title>
	<atom:link href="http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/feed/" rel="self" type="application/rss+xml" />
	<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/</link>
	<description>A slice of lime in the soda</description>
	<lastBuildDate>Thu, 23 May 2013 20:58:00 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.4.2</generator>
	<item>
		<title>By: EllieKim</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-46201</link>
		<dc:creator>EllieKim</dc:creator>
		<pubDate>Tue, 12 Mar 2013 15:26:13 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-46201</guid>
		<description>This and other topics that are relevant for speed traders and institutional investors will be discussed at High-Frequency Trading Leaders Forum 2013 London, next Thursday March 21.</description>
		<content:encoded><![CDATA[<p>This and other topics that are relevant for speed traders and institutional investors will be discussed at High-Frequency Trading Leaders Forum 2013 London, next Thursday March 21.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Curmudgeon</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42257</link>
		<dc:creator>Curmudgeon</dc:creator>
		<pubDate>Thu, 02 Aug 2012 19:26:19 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42257</guid>
		<description>My dissertation research was on intelligent packet routing algorithms for large scale networks.  For this experience and for several famous computing failures in recent history, I have become a fan of the human in the loop.</description>
		<content:encoded><![CDATA[<p>My dissertation research was on intelligent packet routing algorithms for large scale networks.  For this experience and for several famous computing failures in recent history, I have become a fan of the human in the loop.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: jCarl</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42245</link>
		<dc:creator>jCarl</dc:creator>
		<pubDate>Thu, 02 Aug 2012 16:10:07 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42245</guid>
		<description>Complex systems are bound to become unstable. Interconnectivity creates conditions for &quot;Dragon Kings&quot;, Black Swans, Tail Events, however yo want to call them. The phenomenon it&#039;s not  circumscribed to the IT world. We&#039;ve reach criticallity and complexity levels that breed instability.
http://thechinonomist.blogspot.com/2012/04/as-delicate-as-butterfly.html</description>
		<content:encoded><![CDATA[<p>Complex systems are bound to become unstable. Interconnectivity creates conditions for &#8220;Dragon Kings&#8221;, Black Swans, Tail Events, however yo want to call them. The phenomenon it&#8217;s not  circumscribed to the IT world. We&#8217;ve reach criticallity and complexity levels that breed instability.<br />
<a href='http://thechinonomist.blogspot.com/2012/04/as-delicate-as-butterfly.html'>http://thechinonomist.blogspot.com/2012/ 04/as-delicate-as-butterfly.html</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: baroque-quest</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42240</link>
		<dc:creator>baroque-quest</dc:creator>
		<pubDate>Thu, 02 Aug 2012 15:15:45 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42240</guid>
		<description>TFF wrote: &quot;I like the idea of making people responsible for their trades&quot;

I&#039;ll go further.  I want a financial transactions tax, but that is not enough.  If a company crashes the market, I want management to be held personally responsible and not the slap-on-the-wrist nonsense that we see from the SEC.  The corporate veil should not just be pierced, but burned away.  Crashing the market costs a bundle and I want the assets of the company AND management to be used to make people whole.  If that forces management into chapter 7 bankruptcy, too bad, so sad.</description>
		<content:encoded><![CDATA[<p>TFF wrote: &#8220;I like the idea of making people responsible for their trades&#8221;</p>
<p>I&#8217;ll go further.  I want a financial transactions tax, but that is not enough.  If a company crashes the market, I want management to be held personally responsible and not the slap-on-the-wrist nonsense that we see from the SEC.  The corporate veil should not just be pierced, but burned away.  Crashing the market costs a bundle and I want the assets of the company AND management to be used to make people whole.  If that forces management into chapter 7 bankruptcy, too bad, so sad.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: baroque-quest</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42238</link>
		<dc:creator>baroque-quest</dc:creator>
		<pubDate>Thu, 02 Aug 2012 15:03:28 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42238</guid>
		<description>I wrote: “If systems cannot be thoroughly tested due to the nature of the business”
TFF wrote: &quot;Doesn’t that apply in this case?&quot;

I think we are in agreement.  I do not believe that computer trading should exist, period.  These clowns have proven that the nature of the industry is indeterminate.  It is not remotely possible to test all of the variables, not to mention that variables continue to be added.

But then again, I believe that hedge funds and just about every other company on Wall Street are parasites.

You are correct that my Newegg example was not nearly as complicated as computer trading.  A better example would have been Amazon, which has become fairly complex, but things are still sometimes broken.  If people cannot buy on Amazon, no big deal, but if the markets crash, big deal.</description>
		<content:encoded><![CDATA[<p>I wrote: “If systems cannot be thoroughly tested due to the nature of the business”<br />
TFF wrote: &#8220;Doesn’t that apply in this case?&#8221;</p>
<p>I think we are in agreement.  I do not believe that computer trading should exist, period.  These clowns have proven that the nature of the industry is indeterminate.  It is not remotely possible to test all of the variables, not to mention that variables continue to be added.</p>
<p>But then again, I believe that hedge funds and just about every other company on Wall Street are parasites.</p>
<p>You are correct that my Newegg example was not nearly as complicated as computer trading.  A better example would have been Amazon, which has become fairly complex, but things are still sometimes broken.  If people cannot buy on Amazon, no big deal, but if the markets crash, big deal.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: TFF</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42237</link>
		<dc:creator>TFF</dc:creator>
		<pubDate>Thu, 02 Aug 2012 14:53:07 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42237</guid>
		<description>@najdorf, trading once per hour probably doesn&#039;t create any serious issues. But when you are trading dozens of times a MINUTE, the speed of the transactions clearly outpaces any possibility of human control.

I like the idea of making people responsible for their trades (the Flash Crash trades should never have been canceled), but when these issues crop up several times a year I have to question your assertion that &quot;even high-frequency markets can work fine&quot;. Trades that bring down firms add to the systemic instability.

And personally I love volatility. It is very profitable for somebody who trades on fundamentals. Gotta wonder about the chumps who are taking the other side of those trades, though.</description>
		<content:encoded><![CDATA[<p>@najdorf, trading once per hour probably doesn&#8217;t create any serious issues. But when you are trading dozens of times a MINUTE, the speed of the transactions clearly outpaces any possibility of human control.</p>
<p>I like the idea of making people responsible for their trades (the Flash Crash trades should never have been canceled), but when these issues crop up several times a year I have to question your assertion that &#8220;even high-frequency markets can work fine&#8221;. Trades that bring down firms add to the systemic instability.</p>
<p>And personally I love volatility. It is very profitable for somebody who trades on fundamentals. Gotta wonder about the chumps who are taking the other side of those trades, though.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: najdorf</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42236</link>
		<dc:creator>najdorf</dc:creator>
		<pubDate>Thu, 02 Aug 2012 14:46:09 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42236</guid>
		<description>I don&#039;t understand at all why people like the financial transaction tax.  How is trading once per hour more socially undesirable than trading once per month?  Why would we try to regulate markets with such a blunt instrument?  

Instead, let&#039;s make people responsible for their trades.  If you program your computer such that it sells IBM at 0.01, well, looks like you should have programmed your computer better and are going to have to come up with some money.  Secondly, if your behavior causes exchanges to malfunction and be unable to deliver accurate quotes (as in flash crash), let&#039;s target that behavior and ban it.  But on a day-to-day basis, even high-frequency markets can work fine as long as we all appreciate that stocks are going to bounce around and we aren&#039;t entitled to have them smoothly compound every day at a 10% annual rate.</description>
		<content:encoded><![CDATA[<p>I don&#8217;t understand at all why people like the financial transaction tax.  How is trading once per hour more socially undesirable than trading once per month?  Why would we try to regulate markets with such a blunt instrument?  </p>
<p>Instead, let&#8217;s make people responsible for their trades.  If you program your computer such that it sells IBM at 0.01, well, looks like you should have programmed your computer better and are going to have to come up with some money.  Secondly, if your behavior causes exchanges to malfunction and be unable to deliver accurate quotes (as in flash crash), let&#8217;s target that behavior and ban it.  But on a day-to-day basis, even high-frequency markets can work fine as long as we all appreciate that stocks are going to bounce around and we aren&#8217;t entitled to have them smoothly compound every day at a 10% annual rate.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: TFF</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42235</link>
		<dc:creator>TFF</dc:creator>
		<pubDate>Thu, 02 Aug 2012 14:35:52 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42235</guid>
		<description>I do like the idea of a financial transactions tax, though I&#039;m unclear on how that would impact &quot;market makers&quot;? Would the transactions be rewritten to directly match buyers and sellers? Would Official Market Makers be exempt?

Amazing that Knight managed to wipe out 3-4 years of earnings and a third of their equity capital with a single mistake. Somebody won&#039;t be getting a bonus this Christmas! The dollar amount isn&#039;t quite up to the London Whale, but the firm is much smaller.</description>
		<content:encoded><![CDATA[<p>I do like the idea of a financial transactions tax, though I&#8217;m unclear on how that would impact &#8220;market makers&#8221;? Would the transactions be rewritten to directly match buyers and sellers? Would Official Market Makers be exempt?</p>
<p>Amazing that Knight managed to wipe out 3-4 years of earnings and a third of their equity capital with a single mistake. Somebody won&#8217;t be getting a bonus this Christmas! The dollar amount isn&#8217;t quite up to the London Whale, but the firm is much smaller.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: JustinCormack</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42234</link>
		<dc:creator>JustinCormack</dc:creator>
		<pubDate>Thu, 02 Aug 2012 14:29:23 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42234</guid>
		<description>Looks like Knight are now going to go bankrupt or be sold. This will impact the number of people wanting to invest in HFT companies quite a bit with such visible downside.</description>
		<content:encoded><![CDATA[<p>Looks like Knight are now going to go bankrupt or be sold. This will impact the number of people wanting to invest in HFT companies quite a bit with such visible downside.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: SpringTexan</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42233</link>
		<dc:creator>SpringTexan</dc:creator>
		<pubDate>Thu, 02 Aug 2012 14:16:26 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42233</guid>
		<description>A financial transactions tax is indeed the perfect remedy for this and desirable for other reasons as well.</description>
		<content:encoded><![CDATA[<p>A financial transactions tax is indeed the perfect remedy for this and desirable for other reasons as well.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: TFF</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42230</link>
		<dc:creator>TFF</dc:creator>
		<pubDate>Thu, 02 Aug 2012 12:30:47 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42230</guid>
		<description>&quot;If systems cannot be thoroughly tested due to the nature of the business&quot;

Doesn&#039;t that apply in this case?

As for Newegg&#039;s servers, their task is much simpler -- distributing static information that is not time sensitive. It isn&#039;t a big deal if it takes a couple minutes for an update to push through the network (and it might happen more quickly than that).

Might a market be described as a dynamic consensus? A much knottier problem, and one whose behavior depends on all of the participants. Might be able to program something robust if you had a single programming team working with ALL the HFT machines, but they are each doing their own thing (and reacting to code they don&#039;t control). I don&#039;t see how you can possibly pretend to test that!</description>
		<content:encoded><![CDATA[<p>&#8220;If systems cannot be thoroughly tested due to the nature of the business&#8221;</p>
<p>Doesn&#8217;t that apply in this case?</p>
<p>As for Newegg&#8217;s servers, their task is much simpler &#8212; distributing static information that is not time sensitive. It isn&#8217;t a big deal if it takes a couple minutes for an update to push through the network (and it might happen more quickly than that).</p>
<p>Might a market be described as a dynamic consensus? A much knottier problem, and one whose behavior depends on all of the participants. Might be able to program something robust if you had a single programming team working with ALL the HFT machines, but they are each doing their own thing (and reacting to code they don&#8217;t control). I don&#8217;t see how you can possibly pretend to test that!</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: timothydh</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42229</link>
		<dc:creator>timothydh</dc:creator>
		<pubDate>Thu, 02 Aug 2012 12:17:31 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42229</guid>
		<description>Felix, you say &quot;Let’s say we implemented a financial-transactions tax, or moved to a stock market where there was a mini-auction for every stock once per second: I doubt that would cause measurable harm to investors (as opposed to traders). And it would surely make the stock market as a whole less brittle.&quot;  Do you have any concrete data to back up these feelings? Or is it just pure speculation?</description>
		<content:encoded><![CDATA[<p>Felix, you say &#8220;Let’s say we implemented a financial-transactions tax, or moved to a stock market where there was a mini-auction for every stock once per second: I doubt that would cause measurable harm to investors (as opposed to traders). And it would surely make the stock market as a whole less brittle.&#8221;  Do you have any concrete data to back up these feelings? Or is it just pure speculation?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: baroque-quest</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42227</link>
		<dc:creator>baroque-quest</dc:creator>
		<pubDate>Thu, 02 Aug 2012 11:46:38 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42227</guid>
		<description>TFF wrote: &quot;Distributed systems are almost impossible to analyze and can behave in chaotic fashions.&quot;

Nonsense.

If Newegg, a favorite supplier for IT people, were to add servers in a new city to meet demand, prices would not rise or fall.  The only effect would be that availability would be a little better and response times would be a little less.  It is a simple concept known as load balancing.

Yes, the two examples are not completely comparable, but competent people are expected to test their systems before putting them online.  If systems cannot be thoroughly tested due to the nature of the business, perhaps the business should not exist.

The problem with high-speed traders is that they use extremely complex algorithms that depend on a myriad of factors.  The speed of the trade is paramount.  The flash crash and this newest episode has proven that this industry is dangerous.

I agree with TFF regarding his ststement &quot;Hang your hat on unconstrained algorithms and you deserve to lose your shirt from time to time.&quot;  All businesses like the one Salmon described should be shuttered and their equipment given to schools.</description>
		<content:encoded><![CDATA[<p>TFF wrote: &#8220;Distributed systems are almost impossible to analyze and can behave in chaotic fashions.&#8221;</p>
<p>Nonsense.</p>
<p>If Newegg, a favorite supplier for IT people, were to add servers in a new city to meet demand, prices would not rise or fall.  The only effect would be that availability would be a little better and response times would be a little less.  It is a simple concept known as load balancing.</p>
<p>Yes, the two examples are not completely comparable, but competent people are expected to test their systems before putting them online.  If systems cannot be thoroughly tested due to the nature of the business, perhaps the business should not exist.</p>
<p>The problem with high-speed traders is that they use extremely complex algorithms that depend on a myriad of factors.  The speed of the trade is paramount.  The flash crash and this newest episode has proven that this industry is dangerous.</p>
<p>I agree with TFF regarding his ststement &#8220;Hang your hat on unconstrained algorithms and you deserve to lose your shirt from time to time.&#8221;  All businesses like the one Salmon described should be shuttered and their equipment given to schools.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: TFF</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42217</link>
		<dc:creator>TFF</dc:creator>
		<pubDate>Thu, 02 Aug 2012 01:15:48 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42217</guid>
		<description>Distributed systems are almost impossible to analyze and can behave in chaotic fashions. What&#039;s new?

I&#039;m a little less worried about this than Felix, as I don&#039;t see how this will bring down civilization. (They ought to have allowed the Flash Crash trades to stand.) Hang your hat on unconstrained algorithms and you deserve to lose your shirt from time to time.</description>
		<content:encoded><![CDATA[<p>Distributed systems are almost impossible to analyze and can behave in chaotic fashions. What&#8217;s new?</p>
<p>I&#8217;m a little less worried about this than Felix, as I don&#8217;t see how this will bring down civilization. (They ought to have allowed the Flash Crash trades to stand.) Hang your hat on unconstrained algorithms and you deserve to lose your shirt from time to time.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: JRHulls</title>
		<link>http://blogs.reuters.com/felix-salmon/2012/08/01/when-large-scale-complex-it-systems-break/comment-page-1/#comment-42216</link>
		<dc:creator>JRHulls</dc:creator>
		<pubDate>Thu, 02 Aug 2012 00:46:00 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.reuters.com/felix-salmon/?p=16791#comment-42216</guid>
		<description>You might want to look at this interesting post on the Flash Crash as studied by the Lawrence Berkeley National Lab here:  http://oecdinsights.org/2012/06/28/what-can-nascar-teach-nasdaq-about-avoiding-crashes/ which also has links to a scholarly article on the same subject.  I contributed a companion piece on economic models, which comments on the nature of algorithmic trading at http://oecdinsights.org/2012/06/27/going-with-the-flow-can-analog-simulations-make-economics-an-experimental-science/
and the lead article in the series, celebrating Turing, is also of interest. 

This all underscores the question as to the social value of such markets and their unstable hyper-liquidity.</description>
		<content:encoded><![CDATA[<p>You might want to look at this interesting post on the Flash Crash as studied by the Lawrence Berkeley National Lab here:  <a href='http://oecdinsights.org/2012/06/28/what-can-nascar-teach-nasdaq-about-avoiding-crashes/'>http://oecdinsights.org/2012/06/28/what- can-nascar-teach-nasdaq-about-avoiding-c rashes/</a> which also has links to a scholarly article on the same subject.  I contributed a companion piece on economic models, which comments on the nature of algorithmic trading at <a href='http://oecdinsights.org/2012/06/27/going-with-the-flow-can-analog-simulations-make-economics-an-experimental-science/'>http://oecdinsights.org/2012/06/27/going -with-the-flow-can-analog-simulations-ma ke-economics-an-experimental-science/</a><br />
and the lead article in the series, celebrating Turing, is also of interest. </p>
<p>This all underscores the question as to the social value of such markets and their unstable hyper-liquidity.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
