Frequently Asked Questions


Q. What is Deep Learning?

By Nick Brestoff


A. Deep Learning is the "street name" for a multi-layered neural network. There's a lot to learn. Deep Learning practitioners use a Convolutional Neural Network for images. The neural network of choice for text is a Recurrent Neural Network.

In a Forbes article, Why Deep Learning Is Suddenly Changing Your Life (September 28, 2016), Roger Parloff explained Deep Learning in a comprehensive way for images. Andrew Ng, whose long AI resume includes time as a Stanford professor of computer science and AI leadership positions with Google and Baidu) was quoted for his now-famous contention that Deep Learning is the "new electricity." In the next breath, here's what Ng said he meant: “Just as 100 years ago electricity transformed industry after industry, AI will now do the same.” Rather than repeat this great Forbes article, here's the link to it. Just click here.

Q. Wouldn’t the attorneys, or executives, prefer not to know?

By Nick Brestoff


A. With so much going for a litigation early warning system, how could it fail to be a game-changer?[1]  We’ve had more than one person suggest this question: “Wouldn’t the attorneys, or the executives, prefer not to know?” 

Or these variations on that theme:  “What if the enterprise knew and didn’t do anything?  If the failure to act, after having knowledge, were to be discovered, wouldn’t that be worse than not knowing about the threat in the first place?” 

So we must be clear-eyed about this.  Our view is this:

[T]he profession needs to rethink its role from that of an ambulance at the bottom of a cliff (remedial practice) to helping people to manage risks on top of the cliff.  While the practice at the bottom of the cliff can be very profitable, clients and consumers should be reminded to avoid practices that are detrimental in the longer run.

To practise [sic] preventive law, we must first work with relevant data.  Some of our colleagues may not consider this part of the job description of the legal function, but it is down to us to embrace it or watch someone else do so in the course of taking our profession to the next level.  In today’s big data era, this is not an option, but a necessity.[2]

.... (Omitted for brevity.) (Boldface added.)

It seems that there are many reasons “not to know.”   

But we contend that we humans have learned, in other contexts, that it is far better to know than not to know.  The Greeks knew this long ago, and we have already mentioned their phrase for it:  Forewarned is forearmed.  The ostrich defense—sticking one’s head in the sand to avoid knowing about the nearby predator—has never worked very well for the ostrich.  In American jurisprudence, this defense is not well known as a successful strategy and has been alternatively called the dumb CEO defense, dummy defense, idiot defense, or Sergeant Schultz defense.[6]

In our view, it is better to suffer through some number of false positives than to be blind-sided by a preventable litigation catastrophe.

We think Bill Gates would agree.  In 1999, he wrote a book[7] to which he devoted a whole section and six chapters to explain why.  Chapter 10 was entitled, “Bad News Must Travel Fast.”  He even went a step further, and began the section this way:

“I have a natural instinct for hunting down grim news.  If it’s out there, I want to know about it.  The people who work for me have figured this out.  Sometimes I get an e-mail that begins, “in keeping with the dictum that bad news should travel faster than good news, here’s a gem….”[8]        

Gates provides lots of examples, including from the computer industry.  He mentioned IBM, when its mainframe and minicomputer businesses were undermined by the PC; Digital Equipment Corporation, when its minicomputer business was undercut by still smaller PCs, which DEC had dismissed as toys; and Wang, which lost the word processing market when it stuck with putting word processing software on dedicated hardware systems rather than on the PC.[9] 

He also mentions Ford, Douglas Aircraft, and why the United States was not prepared for the attack on Pearl Harbor.[10]

Gates advised this:

“A change in corporate attitude, encouraging and listening to bad news, has to come from the top….  The bearer of bad tidings should be rewarded, not punished….  You can’t turn off the alarm and go back to sleep.  Not if you want your company to survive….[11]

Gates asserts that leaders should heed the early warnings from salespeople, product developers, and customers, but he doesn’t mention the Legal Department.  We can excuse the Legal Departments of the past.  They had no way to see litigation in-the-making, and so could not sound off to give what Gates called an “alert.” 

But now they can see.  An alert from a system intended to prevent litigation is like a smoke alarm: When it goes off, it doesn't necessarily tell you there's a fire; but you have to pay attention to it. (Boldface added.)


[1] The original source of this "answer" is Chapter 24 of Preventing Litigation: An Early Warning System to Get Big Value Out of Big Data (Business Expert Press 2015); see the Book tab.  Intraspexion’s use of this material is permitted by Nick Brestoff's contract with the publisher. Intraspexion’s co-founder, Larry W. Bridgesmith, added the points concerning the criminal sentencing guidelines and Sarbanes Oxley.

[2] Kenneth Tung.  March, 2015.  “A Kodak moment for the legal profession,”, (last accessed April 11, 2015).  In 2008, law practice guru Richard Susskind, who endorsed Preventing Litigation, said as much in The End of Lawyers? Rethinking the Nature of Legal Services (Oxford: Oxford University Press), p. 224.

... (Omitted for brevity.)

[6] U.S. Legal, Inc.  Unknown date.  “Ostrich Defense Law and Legal Definition.” (last accessed April 27, 2015). Furthermore, an ostrich defense does not open the door to favorable treatment for being proactive under the criminal sentencing guidelines, and defies the mandates of Sarbanes Oxley requiring publicly traded companies and their executives to adopt and use proactive policies and procedures to uncover fraud and illegality. 

[7] William H. (Bill) Gates III, with Collins Hemingway.  1999.  Business @ the Speed of Thought:  Using a Digital Nervous System (New York:  Warner Books, Inc. 1999), pp. 159-200.

[8] Ibid. at 159-160.

[9] Ibid. at 179-180.

[10] Ibid. at 180.  For the Pearl Harbor example, Gates cites Gordon Prange, At Dawn We Slept (New York:  McGraw Hill, 1981) at 439-492 (chapters 54 through 59), for communications breakdowns and “fundamental disbelief” on the U.S. side during the weekend of December 6-7, 1941. 

[11] Ibid. at 179.

Q. What's the worst that can happen when attorneys can't see the risks and so try to protect their corporate clients by training their employees how to write?

By Nick Brestoff


A. On February 20, 2013, attorney Lori B. Leskin wrote an article for the American Bar Assn. section on Litigation, addressing the topic of Product Liability. The title: "Assessing Litigation Risks Before It's Too Late." To read her article, click here.

Under a subhead, "Write Right," she wrote:

"Thus, it is important that company employees are trained on how to write and create documents early on in their careers. It is important, too, however, that litigation concerns not serve as a block to true scientific discussion and that employees not be—or feel—discouraged from expressing actual concerns or discussing factual information for fear of litigation. This is not a matter of “covering up” bad information. Rather, the focus in document training is on the creation of accurate documents; it is about injecting thoughtful, common sense discipline into internal corporate writing to avoid the unintended consequences that flow from speculative, inaccurate, or insensitive language. Employees must be taught to think carefully before generating documents, and they must understand the potential harmful consequences their written words can have on the company." (Boldface added.)

Attorney Leskin's advice is carefully phrased. But training employees how to "write right" doesn't preclude other attorneys from trying to train employees to not "write wrong." 

How would anyone train employees to not "write wrong"? The answer, of course, is by giving examples of what not to write. That's what happened in connection with GM's Ignition Switch debacle. GM's attorneys wanted to stop its engineers from using "charged" words, so they created a "word rug" to hide the truth of what its engineers had been saying. In addition to 63 other words and phrases, GM's attorneys advised GM engineers to "not write" the following six:

Challenger, deathtrap, Hindenburg,

Kevorkianesque, Titanic and "you're toast."

Read "These Are The 69 Words GM Employees Were Forbidden From Using" by Sam Frizell, in Time. Published on May 17, 2014. To read Frizell's article, click here

In paragraph 20 of the Consent Decree, GM was required to disavow what GM's attorneys had advised. (Even though most of his article is now behind a pay wall, you can find a link to the Consent Decree in Tom Gara's May 16, 2014 Wall Street Journal blog post article entitled, "The 69 Words You Can't Use at GM." To see the Consent Decree, click on the phrase "legal documents released" in the portion you can still read. Click here for Gara's article.)  

And later, some of GM's corporate attorneys were fired

So that's the worst. 

Q. Do we enable an invasion of privacy?

By Nick Brestoff


Introductory note: Readers are advised that Intraspexion is not entitled to practice law or give legal advice. Readers must NOT rely on any discussion of a legal issue without consulting with their own attorney.  

A. No, not when the appropriate corporate policies in place. We are aware that the act of monitoring internal emails raises a significant privacy issue. While it is not our purpose to provide our clients (or potential clients) with legal advice on this or any other issue, we have considered the subject of privacy in this context at some length.

In Holmes v. Petrovich Dev. Co., LLC, 191 Cal. App. 4th 1047, 1068–71 (2011), the California Court of Appeal held that an employee had no expectation of privacy in the emails she sent to her attorney from a company computer because the company had a policy against using computers for personal reasons and the policy stated that the company could monitor all emails. The court in Holmes emphasized that the computer used to send the emails “belong[ed] to the [company],” that the company had a policy against using its computers for personal reasons, and that the employee was “aware of and agree[d] to these conditions.” Id. at 1068; see also id. at 1068–69 (“Holmes used her employer’s company e-mail account after being warned that it was to be used only for company business, that e-mails were not private, and that the company would randomly and periodically monitor its technology resources to ensure compliance with the policy.”).

But suppose that employee never reads or agrees to the “no privacy” policy or agrees to the policy but someone in authority undermines the policy by conduct; or that the company knows the employee is using a company device, e.g., a smart phone, for personal purposes, pays to use it, and the company does not enforce the policy. In such cases, Holmes may be held not to apply. See, e.g., Mintz v. Mark Bartelstein & Assoc., Inc., 885 F. Supp. 2d 987, 998 (2012).

Q. What software do you use?


A. While some of the software in our patented system (see the tab for "7 Patents") is proprietary and we treat that coding as a "trade secret" and our confidential intellectual property, we are proud to say that portions of our system are powered by dtSearch®. In addition, we use GloVe (open-sourced) and Google's open-sourced version of TensorFlow. TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc. TensorFlow is included in our system under an Apache 2.0 license, the terms of which are incorporated herein in full by this reference.