HN user
Pwnshow https://www.pwnshow.com/
While user count can affect absolute disclosure volume it does not drive the main comparison in my paper. What I do is to use Xen advisories to measure upstream dependence: how much of the Qubes public advisory record is attributable to Xen/CPU/upstream components rather than Qubes core logic? That attribution result is internal to the Qubes record, so it is much less sensitive to Xen’s much larger user base.
BTW, project summary is here https://www.pwnshow.com/investigations/INV-006/
Author of the paper here; AMA.
Yes, this is a tactic some companies resort to in order to incentive resignations. However, as you correctly highlight, this typically results in a loss of talents. AFAIK, Intel is already experiencing challenging times. Attracting – and retaining – new talents is especially hard to do, when the job marketplace learns about this peculiar work culture that prioritises paying dividents to talent retention. I would be very happy to be proven mistaken, for the sake of friends working at Intel.
No. Employees are neither VCs, nor credit institutions, nor social benefactors, nor non-profit, nor pro-bono volunteers. They work for a wage – And, yes, first hired that joined the ranks of executives with a vetted interest in the company (e.g. stock options) are already something other. Paying the wage is an obligation, if the employers want the work to be provided. Risks – and the associated rewards - are primarily with the investors. A work culture that shifts the risks onto workers has to be rethought.
If I were an Intel employee I would never accept a pay cut "justified" by the company's willingness to pay dividends. Investors get dividends when the management creates a reasonable amount of profits, not when employees have to waive their dues to do investors a favor.
Interesting op-ed. I support the remark about how the scientific community made science a team sport – for no good scientific reason.
However the claim "[the approach] was flawed in a way that undermined its efficacy and resulted in thousands if not millions of preventable deaths" is not demonstrated in the article. Conversely there were countries (e.g. Italy) where the epidemiological distribution of SARS-CoV-2 and the determinants (e.g. risk factors such as young and old people living together in the same households) were costing lives, that if regarded in the thousands would be grossly underestimated.
But a sequence of rapatronic stills would be very interesting.
You have a short sequence of stills in the Wyckoff's footage linked somewhere else here. Rapatronic cameras – Wyckoff explains – where typically mounted in a rack of six, where each captured a single frame.
A back of envelope calculation for the shutter speed of a Pimoroni LCD shutter [1] suggests 1/15s is the fastest speed available. [2]
[1] Pimoroni LCD shutter [Not available any longer ] https://shop.pimoroni.com/products/lcd-shutter
[2] Testing the amazing Pimoroni LCD shutter https://youtu.be/i3GCfAtdZYI
Charles Wyckoff, who worked for EG&G (Edgerton, Germeshausen and Grier), showcasing a rapatronic camera and its peculiar shutter. [1]
Well, this is something that happens pretty much often in a number of verticals. Just think about how cars, even the most sophisticated, look when compared to the prototypes exhibited by the their manufacturers years before at the car shows. Manufacturing is also about costs/margin tradeoffs.
That being said, having had the chance to tinker with those machines in the 1980s I can tell you that I owe them, and to my father, very much about my obsession for computer science.
In the early 1980s my father had in his office a couple of these Olivetti BCS 2025 [1], which share with the Bellini's TCV 250 some design ideas, such as the metal plane encapsulating the keyboard.
[1] http://museo.dagomari.prato.it/singolo.php?cod=149&ord=1
In a nutshell:
[Researchers] discovered something new – a multicellular self-assembly process in E. coli. Researchers observed unattached, single-celled organisms combining into four-cell rosettes, a natural multicellular formation thought to be uncommon in bacteria.
When I request new imagery my expectation is to have the copyright of the resulting image. Whilst I would have the need for high-resolution satellite images, I am not interested in acquiring licenses of any of them.
Probably most businesses will not charge any interaction with ChatGPT per se. They will either bill more abstract services (e.g., booking service, where ChatGPT will be hiding under hood of a support agent) or down to earth goods.
Whilst ChatGPT is built on a fine-tuned GPT-3.5 model, I was wondering whether it will be possible to further fine-tune ChatGPT models on Azure. There are quite a few ideas I would like to tinker with.
The reality is that you cannot trust that your machines are not compromised.
The only option we are left with is to operate under the assumption that, indeed, our machines are permanently compromised.
Absolutely, the same can be said about C – And how many succeeded in doing so in the latest 30 years of Web history?
I am reminded about a quote attributed to Keith Martin (mathematician): "Theory is important, at least in theory"
Okay. But Fredric Jameson establishes that in postmodernism we have experienced a weakening sense of historisity such that what is, what was, and what will be all exist as presents in time. 1970, 1991, 1992, and 2017 all happen simultaneously.
Okay with the postmodernism, but relying on a memory unsafe language to implement a server-side web framework in 1970, 1991, 1992, and 2017 is equally anachronistic. That being said much love to Forth! – Or, as the post-modern philosopher Slavoj Žižek is used to say: "and so on and so Forth".
Seconded.
in the next 5 years
I will not hold my breath. I tested the first (admittedly rudimentary) HMDs in the mid 1990s. VR was f-ing hot back in the days among those of us interested in "computer graphics". According to commentators, VR would have disrupted our lives over the next a few years. Here we are.
I have not seen mentioned so far in this thread the following book, which I can't recommend more highly:
Understanding Machine Learning: From Theory To Algorithms – Shai Shalev-Shwartz
Sure. Publishing means having your name listed among the authors. This does not necessarily translate in the production of results, or in the active participation in the writing process.
An interesting statistic would be the number of scientists who are able to publish a paper every five days or so, while going solo in their work. Essentially nobody, I suppose, for so many reasons I will not list here.
Congrats, this is so handy (pun intended)! :-)
Please consider adding a custom watch face to the Coros watches such as the Vertix.
Also consider that there are timezones offset by 30 mins.
There is a line of research according to which social media create addiction [0]. When we are addicted to social media, we have less and less time for tinkering and pushing the envelope, because a sizeable part of our time is claimed by such media.
[0] Griffiths, M. D. (2012). Facebook addiction: Concerns, criticism, and recommendations: A response to Andreassen and colleagues. Psychological Reports, 110, 518–520. https://doi.org/10.2466/01.07.18.PR0.110.2.518-520
Sun Microsystems could not have a more special place in my hacker heart. In my youth Sun was all over the place; I was dreaming working for them – as I would eventually did later in my life – and, admittedly, I would have done so even only on the basis of their uber cool logo. I remember working on their Sun Ray thin clients for a military contract. Good times. Thanks for sharing this video.
Not a big deal. You can edit a PDF programmatically, a la pdfcpu [0] or similar.
In my opinion this could turn in a top seller – and I wish you so – if only you could add the means to sync Hyperpaper Planner to other calendars, such as Apple's and Google's ones.
I would typically need to quickly import calendar files related to the various online events I have to take part. And the planner should account for the differences in time zones, so that I do not have to.
That of which is actually provided are 125M partial pre-images.
A cryptographic hash function – SHA-256 included – is not inverted, because any "good" cryptographic hash function has probably infinitely many pre-images for any given hash value, if the input space is unbounded (i.e., {0,1}*).
It would be nice to have vanity URLs for the profiles. UUIDs are a no go in this application scenario.