In RISKS-24.19, there were three reports about the College Board reporting problems with SAT scoring. First, the College Board said that about 4000 tests were misgraded, with results off by no more than 100 points. Then the College Board admitted some were off as much as 200, or maybe even 400 points (out of 2400 total). Today, the College Board admits that there were an additional 27,000 score sheets that weren't rechecked, and they found 375 more students who received incorrect scores. There was no disclosure of how far off these results were. The article notes "The College Board said that from now on all answer sheets would be scanned twice, among other new precautions, and that it would retain consulting firm Booz Allen Hamilton to perform a comprehensive review within 90 days." (http://www.cnn.com/2006/EDUCATION/03/23/sat.scoring.error.ap/index.html) Two things struck me about this sequence of revelations: (1) Does the College Board even have a *legal* obligation to disclose this information? Could it be that this has happened in the past, and without the increased scrutiny caused by the disclosures of personal information leakage, they might never have told the students or the public? (2) On the positive side, it's a good thing there's paper to double-check. If these were votes on paperless DREs instead of SAT scores, there would be no way of knowing that they had been miscounted. As a parent whose oldest child went through the process last year, I'm relieved that she's not having to deal with this headache - and I feel sorry for any student who made decisions on where to apply based on SAT scores. (I know we used my daughter's scores to help find target schools - if they had been off by a few hundred points, she might not even have applied to the school she selected.) While colleges can reexamine the applications in light of corrected SAT scores, there's nothing that can be done for applications that weren't submitted based on incorrect results. Karen W. Arenson in *The New York Times* today is reporting that the College Board has now admitted that the maximum error was 450 points (out of 2400). The College Board had previously claimed 100, then 200, then 400. Her article included this wonderful quote: "Everybody appears to be telling half-truths, and that erodes confidence in the College Board," said Bruce J. Poch, vice president and dean of admissions at Pomona College in Claremont, Calif. "It looks like they hired the people who used to do the books for Enron. My next question is what other surprise we're going to hear about next." http://www.nytimes.com/2006/03/23/education/23sat.html?_r=1&oref=slogin [Lauren Weinstein noted that in its statement, the Board said Pearson would ensure that all answer sheets were "acclimatized before scanning" and would scan each answer sheet twice. Pearson will also improve its software to detect whether answer sheets have expanded because of humidity. PGN] [Jeremy's point about the paperless DREs is apt, but this case reminds us once again that even paperfull media such as optical scanning can have serious problems that require oversight and the willingness to perform meaningful recounts -- which are of course impossible with the current breed of paperless DREs. PGN]
Court-at-law recount suspended; Electronic machines not providing all info Paul A. Anthony, 21 Mar 2006 On orders from the Texas Secretary of State's office, the recount for the Tom Green County Court-at-Law No. 2 race has been suspended midway through its second day. About 1:30 p.m. today, county Republican Chairman Dennis McKerley stopped the recount after workers found discrepancies of as much as 20 percent between what was counted Monday and what was reported Election Night. "We're having some trouble with the electronic equipment," McKerley said. Apparently, McKerley said, new electronic voting machines provided by vendor Hart InterCivic are not printing ballots for every vote cast on the machines. During recounts, which must be done by hand, the machines are designed to print out separate ballots for every vote. http://www.sanangelostandardtimes.com/sast/news_local/article/0,1897,SAST_4956_4559073,00.html
"A baby boy died after an untrained doctor pressed the wrong button on his bypass machine because it was a less `horrid' colour than the other, an inquest heard yesterday. ... [The doctor] was unaware how to use the machinery, as were most of the team." http://www.telegraph.co.uk/news/main.jhtml?xml=/news/2006/03/21/nhs121.xml
This one seems to impact Hewlett Packard employees in the U.S. - I do not know if those in Canada and elsewhere in the world are impacted. No word on use of encryption to protect the data, so I suspect it was NOT protected at all. Will they ever learn? A laptop computer belonging to Fidelity Investments and containing sensitive data on about 196,000 retirement-account customers was stolen last week, the company said. http://www.usatoday.com/money/industries/brokerage/2006-03-23-fidelity_x.htm
Evidently Fidelity lost a laptop containing the HP retirement records. No explanation why it was reasonably on said laptop. To their credit, they sent UPS letters to everyone, but: a) The letters contain an 800 number to call b) The 800 number wants you to key in your social security number before talking to a person. Well that is not a very good design! At least the folks at the main Fidelity number knew how to confirm the special number. I was calling to tell them I got someone else's letter at my address in addition to my own, but I was seriously surprised by the "enter ssn". I would note: - Poor security practices (data on laptops) - Inability to learn from other companies previous misfortunes + An apparently serious response - A poorly designed response - Bad database data - will I get this fellow's pension too?
Here's another example of problems with automated language processing. http://www.wired.com/news/wireservice/0,70453-0.html?tw=rss.technology Amazon Changes 'Abortion' Queries Amazon.com said Monday it had modified the way its search engine handles queries for the term "abortion" after receiving an e-mail complaint that the results appeared biased. Until the recent change, a user who visited the Seattle Internet retailer and typed in the word "abortion" received a prompt asking, "Did you mean adoption?" followed by search results for "abortion." Spokeswoman Patty Smith said the automated prompt was purely based on technology, and that no human had made the decision to show the question. "Adoption and abortion are the same except for two keystrokes," Smith said. "They also, in this case, happen to be somewhat related terms." Still, Smith said she and other company officials decided to remove the question after receiving an e-mail complaint and deciding that it raised a valid concern. People who type in the term "adoption" do not see a prompt asking "Do you mean abortion?"
If you're concerned about privacy, you may be worried about merely throwing away those preapproved credit card applications that come in the mail, especially when they're pre-filled with your name and other information. Indeed, the Federal Trade Commission and many banks recommend tearing up those applications before discarding them. But Rob Cockerham, my favorite empiricist, decided to test how well that strategy actually works. He tore up an application, taped it back together, and mailed it in. Did the bank process the application and issue him a card anyway? One guess. http://www.cockeyed.com/citizen/creditcard/application.shtml
> He tore up an application, taped it back together, and mailed it in. Ahem. He tore up an application, taped it back together, filled it out *with a change of address requested*, and mailed it in.
The spin on the story in RISKS-24.20 was "how awful that a judge says it's illegal to use a secure delete program." But how is this different from a disgruntled employee shredding the only copy of paper files of valuable customer information before quitting to start his own business in competition with his former employer? It should make a difference, of course, whether the deleted files were valuable to the company, and if they were the only copy of the information. The ex-employee made the additional argument that his employment contract specified that he was to return or destroy data upon leaving the company. The company asserted that he had broken the contract and so those the authorization implied by those terms were no longer in force. But the story reports that this was an appeals case. Based on the story, it appears that the judge did not say that files were deleted illegally, only ruled there could be facts in the case which would cause the deletions to be considered as damage and unauthorized. The case was sent back to the lower court so that these facts could be determined. Sidney Markowitz http://www.sidney.com
Here's Microsoft's own description of Excel from online book that came with my copy of the software: > Microsoft® Excel 2004 for Mac® > Use this analysis and spreadsheet program to evaluate, calculate, and > analyze data. Make use of the improved charting and page layout > capabilities to illustrate your data and make it look good in print. An "analysis" program, designed to "analyze data". No mention of accounting. For a scientist, to "analyze data" involves computing statistical summaries and plotting, not silent conversions of data labels. Furthermore, I don't think the work in question used Excel as a database program, but rather as a program to analyze the results of microarray experiments. This task is entirely within the job description for Excel quoted above.
Re: Deltuvia Actually, if you follow the references in the original article, both bioinformatics programs are written in Java with SQL back-ends. "Tab-delimited file suitable for loading into spreadsheet programs" is one of their listed output options. So the problem was introduced by the authors of the original report when they decided to load those output files into Excel for viewing. Re: McCormick" > ... it will often ignore the double-quotes that are intended to distinguish character from numeric fields. Yes, it does that. Note, however, that there is no standard for CSV format. Some applications allow special characters (such as newlines: record separator) inside double-quoted values, some don't. Some applications escape a double quote inside double-quoted values with a backslash (C-style), some use a second quote (SQL-style), some simply can't handle it. There is no way to disambiguate non-text values, such as 20060318. MySQL outputs null fields as ",\N," whereas most others do just ",,". And so on. Which is not as bad as tab-delimited files (output of the two bioinformatics programs in question) where on top of all of the above, a single tab may replaced by 8 (or some other number) of consecutive spaces and there is an option to "not treat consecutive spaces as one". (I.e. to treat "\t\t" as a null field.) Of course, to most parsers a whitespace is just a whitespace, be it "\t" or a " ", so the end result is you get 8 extra null columns because you previously looked at the file in some helpful text editor that quietly replaced tabs with spaces for your viewing pleasure.
My company often gives clients data in CSV formated file that doesn't end in .CSV. This data is usually imported into an accounting system but sometimes users want to look it over in Excel (if it isn't in Excel it isn't data to some people) so they open Excel and then open the file thus bringing up the "Text Import Wizard". The wizard is pretty straight forward, you select delimited then select comma as your delimiter and click Finish. Here is the catch; Excel brings all the columns in using the "General" format, not the "Text" format unless you specify this on the last screen (3 of 3) of the wizard which is often skipped. Thus data that starts with a zero or has a lone 'E' with numbers is often mis-represented. You would think that data brought in via a TEXT Import Wizard would be treated as text but unfortunately this is not the case.
While working on a joint UK / German product development we discovered that the 'standard' separator employed in many German CSV files is the semi-colon ';' - I do not know why. This property is defined in the Regional and Language Options of the Machine as described in the Microsoft Excel Help (in case anyone should need it) : Change the separator in a CSV text file 1. Click the Windows Start menu. 2. Click Control Panel. 3. Open the Regional and Language Options dialog box. 4. Click the Regional Options Tab. 5. Click Customize. 6. Type a new separator in the List separator box. 7. Click OK twice. Note After you change the list separator character for your machine, all applications will use the new character. You can change the character back to the original character by using the same procedure. Naturally, on my machine (Windows 2000) the above 'Help' was found like this: Change the separator in a CSV text file 1. Click the Windows Start menu. 2. Click Control Panel. 3. Open the Regional Options dialog box. 4. Click the Numbers Tab. 5. Click Customize. 6. Type a new separator in the List separator box. 7. Click OK twice once.
For example, the RIKEN identifier "2310009E13" was converted irreversibly to the floating-point number "2.31E+13." That should have been 2.31E+19. Error of the original author, or even further error of Excel? (the original page doesn't seem to offer access to the e-mail addresses; I had wanted to copy the authors too) Olaf 'Rhialto' Seibert rhialto/at/xs4all.nl
Copyright Gone Mad (copyright Robert M. Slade, 2006) (with that little (c) symbol thrown in for good measure) I got asked to do a 20-year retrospective on computer viruses for a tech magazine. There were a few oddities about the request, such as a demand for graphics. I normally don't do graphics, but I had such a fun time doing the article that I gave in, and finally put together quite a piece, I thought. It was a gas going back over all the stuff I've seen over the years. You may never see it. See, I got this phone call from the magazine today. It seems that some of the wording in my article bears a striking resemblance to a site on the Internet: "Robert Slade's Computer Virus History" at http://www.cknow.com/vtutor/RobertSladesComputerVirus.html. This is surprising? I've been writing articles, series, and books about viruses since the darn things started. As a matter of fact, it's a bit surprising that they didn't find more sites with my stuff on it, especially since there have been dozens of examples that I've seen myself, over the years, where people have used my material and passed it off as their own. But it seems that this outfit has a policy where they won't publish anything that has already appeared on the net. I suppose that's fair enough. Everybody is getting really antsy about copyright violations these days, and, as somebody who does an awful lot of writing, I suppose I should approve. Except I don't. The crackdown (and crankdown) on copyright and copying is making it hard for a lot of us who are relying on our own research and writing. After all, who else am I going to use for material on virus history? Oh, lots of people were there, but who else wrote it down? I do go back (and did go back, for this article) and check on specifics, and even made corrections on items we've found out more about. But, by and large, if I want to generate a decent timeline of what happened, I have to rely very heavily on my own stuff. Except, now I can't. Well, like I said, you may not get to see the history article. Or, if they are willing to bend their policy a bit, you might. But I'm willing to bet that their policy is more important to them. After all, they can always get another writer to do it for them. Of course, in all probability he won't know anything about the history of viruses. Or, he can read my stuff. And reuse it. copyright Robert M. Slade, 2006 (with that little (c) symbol thrown in for good measure) email@example.com firstname.lastname@example.org email@example.com http://victoria.tc.ca/techrev or http://sun.soci.niu.edu/~rslade [Ironic. I keep Robert's copyright line in his reviews, despite the RISKS info file that once upon a time said that by default everything that appears in RISKS is fair game if used with appropriate credits. I just discovered that the relevant wording in the risksinfo file somehow got deleted somewhen along the way, and I suppose I'd better fix that. Or perhaps it is better to leave it unspecified so that others can quote Robert without his permission! PGN]
OSDI '06 Call for Papers [Adapted for RISKS by PGN] 7th Symposium on Operating Systems Design and Implementation (OSDI '06) Seattle, WA, USA, November 6-8, 2006, Sponsored by USENIX, in cooperation with ACM SIGOPS http://www.usenix.org/events/osdi06/cfp/ The seventh OSDI seeks to present innovative, exciting work in the systems area ... on the design, implementation, and implications of systems software. The OSDI Symposium emphasizes both innovative research and quantified or illuminating experience. OSDI takes a broad view of the systems area and solicits contributions from many fields of systems practice, including, but not limited to, operating systems, file and storage systems, distributed systems, mobile systems, secure systems, embedded systems, networking as it relates to operating systems, and the interaction of hardware and software development. We particularly encourage contributions containing highly original ideas, new approaches, and/or groundbreaking results. [Full papers are due by 24 Apr 2006.]
Team Software Process Symposium 18-20 Sep 2006, Omni Hotel, San Diego, California Web: http://www.sei.cmu/edu/tsp/symposium.html Theme: Measurable Improvements in Team Performance Deadline for abstracts 28 Apr 2006 The first Team Software Process (TSP) Symposium will include all yearly TSP activities. The conference will bring together users, adopters, and developers of the TSP, those involved in its development and transition, and those who are new to the technology and eager to learn more. Attendees will have the opportunity to exchange ideas, concepts, and lessons learned concerning the experiences, best practices, and suggested introduction strategy for the TSP methods and practices. **** All inquiries to firstname.lastname@example.org **** Jodie Spielvogle, TSP Team Software Engineering Institute, 4500 Fifth Avenue, Pittsburgh, PA 15213 Phone: 412 / 268-6504 FAX: 412 / 268-5758 E-mail: email@example.com
BKNTSCTL.RVW 20051204 "Network Security Tools", Nitesh Dhanjani/Justin Clarke, 2005, 0-596-00794-9, U$34.95/C$48.95 %A Nitesh Dhanjani %A Justin Clarke %C 103 Morris Street, Suite A, Sebastopol, CA 95472 %D 2005 %G 0-596-00794-9 %I O'Reilly & Associates, Inc. %O U$34.95/C$48.95 800-998-9938 fax: 707-829-0104 firstname.lastname@example.org %O http://www.amazon.com/exec/obidos/ASIN/0596007949/robsladesinterne http://www.amazon.co.uk/exec/obidos/ASIN/0596007949/robsladesinte-21 %O http://www.amazon.ca/exec/obidos/ASIN/0596007949/robsladesin03-20 %O Audience a- Tech 2 Writing 1 (see revfaq.htm for explanation) %P 324 p. %T "Network Security Tools" The preface states that the audience for the book is comprised of anyone who wants to program their own vulnerability scanners, or extend those already available. It assumes familiarity with six of the major tools in that class, as well as Perl. Chapter one deals with writing plug-ins for Nessus. It covers the installation and quick use of the program, and then outlines the Nessus Attack Scripting Language, including a few sample scripts. The Ettercap network analyzer and its plug-ins (in the C language) are in chapter two. (An overview of authentication for the ftp protocol is provided in order to discuss looking for ftp passwords.) The Hydra password sniffer (and SMTP authentication) is described in chapter three, as well as the Nmap port scanner. Chapter four looks at plug-ins (in Perl) for the Nikto Web scanner. The Metasploit Framework generic exploit development platform is examined in chapter five, which also has a brief explanation of stack overflows. Chapter six discusses analysis of (mostly source) code for Web applications in a search for vulnerabilities, reviewing the PMD Java analysis tool, and reprinting pages of Java source code. Part two turns to writing network security tools. Chapter seven is primarily a tutorial on Linux kernel modules. Using Perl to write a Web application scanner is in chapter eight. SQL injection, and testing for error message responses, is examined in chapter nine. Chapter ten covers the use of the libpcap library for producing network sniffing utilities. Packet injection, using the libnet library and AirJack device driver, is in chapter eleven. While a lot of sample code is given in this text, ultimately it is about using a bunch of tools. The examples and exploits are interesting, and do provide an indication of limited types of testing utilities that could be developed. copyright Robert M. Slade, 2005 BKNTSCTL.RVW 20051204 email@example.com firstname.lastname@example.org email@example.com http://victoria.tc.ca/techrev or http://sun.soci.niu.edu/~rslade
Please report problems with the web pages to the maintainer