Forgot your password?

Comment: Re:Read Slashdot (Score 1) 479

by larwe (#47982347) Attached to: Ask Slashdot: Finding a Job After Completing Computer Science Ph.D?
The PhDs I worked with didn't know how to do what they were hired to do, because they lied about their abilities to actually do anything practical. Let me take one really simple example: PhD is tasked with implementing over-the-air firmware downloads for a product (there's complexity there, I know it sounds like an intern job, but there really was work to do). Said PhD was told to verify the CRC of the downloaded file for integrity purposes before attempting to flash the device. How did she do this? She downloaded the file (note: said download included, free of charge, a meta-file including a CRC of the known-good data on the server). She CRC'd the data she'd downloade, then CRC'd it again, and proceeded if those two CRCs matched. This is the kind of bullshit I've had to deal with when working with PhDs. Everything practical is left as an exercise for the reader - the PhD is focused on his/her higher calling, or whatever. As a manager of a team that needs to deliver practical implementations of well-understood technology, I don't need that. See also: Dunning-Kruger effect, which I have seen vigorously at work in the minds of PhDs.

Comment: Re:Read Slashdot (Score 1) 479

by larwe (#47978997) Attached to: Ask Slashdot: Finding a Job After Completing Computer Science Ph.D?
I can't agree more with this commenter. After having to live with bad hiring decisions, and then making a couple of my own, I have made a firm resolution not even to consider a candidate with a PhD. That's an instant decline, right there. *EVERY* single PhD I've worked with has been an insufferably bad organizational fit, and almost all of them have ultimately been terminated (after being on a PIP). The remainder have been moved out into satellite offices to work on sinecure projects. In fact, I was musing about this just the other day after another incident, to the point where I was going to post to my social media accounts a call for someone - anyone - who works in a real job (i.e. making a product for profit, not working at an academic institution) - to pipe up and tell me that they have a PhD on their team who is useful and effective. As for your comment about your PhD being in an esoteric and useless field - this is essentially a tautology.

Comment: Better yet, store it "nowhere" (Score 1) 502

by larwe (#47581249) Attached to: Judge: US Search Warrants Apply To Overseas Computers
Better still, web-based companies with datacenters in many jurisdictions could store your data in a completely distributed fashion, where it isn't possible to retrieve the original without access to all (or at least several) of the servers. So they could subpoena all the data held in the US, and the UK, and Australia, and all the other surveillance states, but without a copy of the complementary data in Switzerland and Belize and on a pirate barge in the middle of the Pacific, they won't be able to reconstruct what you actually have stored up there. Better yet, if they contract with third-party storage providers wholly resident in those other countries, the US (for example) would have zero basis to subpoena those other companies - they are not "doing business" in the United States.

Comment: Re:Can a company patent it? (Score 2) 207

by larwe (#47256843) Attached to: Century-Old Drug Reverses Signs of Autism In Mice
Off-label use of a drug currently approved for a given clinical use is very easy to achieve. However, looking at the side effects of this pharmafossil, it's looking like the side effects are worse than the condition it is being proposed to treat. I can't see an adult consenting to take it. Parents could force their children to be dosed with it, I guess.

Comment: Re:Makes forensic avoidance simple (Score 1) 145

by larwe (#47096023) Attached to: The Internet Is Now Part of the Crime Scene
Completely missing my point. Assume that datamining a la NSA, FBI, etc, actually worked to protect the citizenry from thoughtcrime before it becomes realcrime. Thus, the best way to avoid that worm, if you wanted to commit a crime, would be to delete all your social media accounts so they can't even get a network diagram for you, let alone posting bullshit rants about whatever you intend to use as your justification for going crazy and killing people. Of course, the actual fact is that predictive analysis like this is worthless, because humans are non-deterministic creatures.

Comment: Re:I work at exactly such a startup (Score 1) 274

by larwe (#46920665) Attached to: Ask Slashdot: Joining a Startup As an Older Programmer?
I've wanted to go for a VERY long time. And they're getting close to finishing the final sarcophagus that will go over the NPP. I want to see it before that happens. Of course I'm mainly going for Pripyat - I've seen the awesome pictures, I want to see it in person. I miss the 1980s.

Comment: I work at exactly such a startup (Score 2) 274

by larwe (#46914805) Attached to: Ask Slashdot: Joining a Startup As an Older Programmer?
I'm working at a similar second-stage startup (has had significant second/third round investment but still very young). I'm turning 40 this year [yay birthday vacation to Chernobyl!], there are a lot of young'ns in the office. I came from a Fortune 500 company full of processes and requirements and paperwork and ... NEWSFLASH: NOBODY CARES. We are a startup. Many of the MVPs are actually remote. They may not be wearing pants on any given day; I'm not sure. One of the reasons I was hired was to add maturity/realworld regulatory-compliant experience to the company. I have about 7 people in the USA reporting to me, and a few more than that in different countries. Average age of my team is probably mid 20s, but it ranges from "just turned 21" to "early 50s". I need all of those people, for different reasons. Be excellent. That's why they're going to hire you.

Comment: Re:They're not going to get better results... (Score 4, Insightful) 110

by larwe (#46886479) Attached to: SpaceX Looking For Help With "Landing" Video
There is so much wrong in this I barely know where to begin. Nobody records analog streams for digital TV data, but this is completely irrelevant. Everybody records analog streams for spacecraft telemetry because you can't post-analyze an improperly demodulated digital data recording. Doppler velocity measurement is also performed from the raw signal, for example by mixing with a signal at the original (TXCO-controlled) carrier frequency and observing the beat. The A/D stage is NOT demodulation, it's digitization. A digitized recording of an analog waveform is nothing even remotely akin to recording the binary output of a demodulator stage. (For practical purposes, a modern recording would, indeed, be digital - but it would be a digital recording of the original received waveform, not simply a recording of the realtime output from a demod). Having the original waveform to look at allows different types of filters to be tried and applied, not just the single set of parameters that were in the realtime decoder on board the aircraft. It's absolutely the baseline for data recovery in this type of application. Telemetry = "remote measurement", a term used to refer to data streams downlinked from the vehicle that are not crew communications or control data, which includes still and moving image data.

Comment: They're not going to get better results... (Score 4, Interesting) 110

by larwe (#46886141) Attached to: SpaceX Looking For Help With "Landing" Video
... until they post an analog recording of the telemetry. The bitstream *as decoded* is corrupted because of demodulation errors, and you can't reconstruct data that isn't there. If they have an analog recording, then analog filters can be applied to that in an attempt to create a cleaner input signal to the demodulator stage. An analogy: They have taken a picture of a page of text, rather out of focus and dark, and used OCR software on it. All they are giving us is the output of the OCR software. We need to see the original picture so we can apply better filtering/contrast adjustments to it before attempting pattern recognition.

Comment: Re:tie that to K'nect camera (Score 1) 108

by larwe (#46762121) Attached to: 52 Million Photos In FBI's Face Recognition Database By Next Year
Those won't work, there are acceptance standards for those photographs. No headgear (possible religious exemption), no tinted eyeglasses, etc etc. Also, how many women do you know with a full mustache and knee-length beard? (I realize the answer to this may be nonzero, but it's going to be small). When I went to renew my passport a few years ago [Australian], they had additional requirements "neutral expression, no smiling" and they were explicit about the fact that this was to improve facial recognition DB matching. All this is nothing at all compared to the databases in the United Peoples' Democratic Republic of Europistan, of course. Interestingly, all this facial recognition and cross-referencing is a real problem for spies. Passports with biometric information in them that can be cross-referenced to a central database are a serious problem to a guy whose job is to enter Russia as Mr. John Smith, tourist today and enter it again next week as Mr. Alphonse Gambolputty, international financier.

As in certain cults it is possible to kill a process if you know its true name. -- Ken Thompson and Dennis M. Ritchie