771 karma · joined September 14, 2010
I apply machine learning to problems in chemical and pharmaceutical industries.
edit: Minor increase in clarity, re:reply
MIT actually expressly opposes in this document the Keker & Van Nest LLP proposed modification of the protective order, without redaction. And in fact brings up a serious procedural misconduct of Keker & Van Nest LLP (they should have already destroyed any copies they possessed already because the criminal case was fully concluded).
Apparently JSTOR also supports redaction: " JSTOR’s response was similar to MIT’s. “We believe the information we provided to the United States Attorney’s Office in this case should be made open and available to the public. In a letter dated Feb. 25, we agreed to the lifting of the protective order so long as the articles downloaded from JSTOR were not released and the identities of our staff are protected,” said Heidi McGregor, a JSTOR spokeswoman, in an email to The Tech. “We do not agree that individuals’ names need to be included with these materials to serve the public interest.” "
" “Although the United States and representatives of Mr. Swartz agreed on many proposed modifications to the order, the United States and Mr. Swartz’s representatives did not reach agreement on the scope of the redactions,” Pirozzolo said.
“The United States expects to respond to the motion within the time provided by the district court rules,” he said. “It will also request that individuals potentially affected by the modification of the order be given an opportunity to be heard on the proposed modifications.” "
Given that last part you can certainly expect the MIT documents to come out much much faster than Gov documents.
The blog author saying there is no clear trend in his random blob of points is misleading. The trend is there, it is just difficult to tell the density of points in his scatter plot because of the marker size and plot size.
MIT does not and should not decide prosecution strategy. This was not a civil case, but a criminal prosecution in the American tradition of criminal justice. The victim is not the one prosecuting the crime but rather the government prosecutes on behalf of all people, not on behalf of the victim exclusively.
http://www.nature.com/srep/2012/121220/srep01001/full/srep01...
First one came out of MIT http://web.mit.edu/newsoffice/2012/infrared-photovoltaic-062...
If a student lobs their id over the fence, then walks out of the campus is the system going to think they were at school all night?
Seems very problematic to implement an EZ-pass style system.
How many apps went down? Did you violate iTunes/App Store T+C in some noticeable way? Finally, did you violate someone elses rights copyrights, trademarks, intellectual property such that they may have sent a demand letter to Apple?
Perhaps in retrospect, having all the apps tied to a single entity was a bad service for your clients representing a single point of failure.
I'd put up a pdf of a generated assignment that can be scan/fax graded online right away in an easy to access fashion. That is what I most wanted to take a look at, and it looks like the demo won't render the document.
Do you put a QR code on each assignment to uniquely identify a student? So papers have to be handed out precisely to each student? Or do you have students bubble in names/identification numbers?
How are you going to handle the kind of problems that would be created by using this technology in a class of <8th graders? Crumpled/dog eared papers, intentional attempts to make forms hard to computer read. Have you thought about linking to Amazon's Mechanical Turk to have human grading of difficult to parse entries?
Is your product going to be crushed by computers/tablets for every student in future classroom environments? If assignments are all digital (i.e some college textbook environments like 'Mastering Physics') the scan to grade edge slips out. Especially when providers can offer prewritten educational material questions/exams/content as part of their platform.