Profile cover photo
Profile photo
Anton Burtsev
Anton's posts

Post has attachment

Post has attachment

Post has shared content
MIT professor +Seth Mnookin pens a gripping essay for +The New Yorker on the discovery of NGLY1 deficiency, my son's disease.

It's a good summary of what I've been doing for the past six and a half years.

It's also a glimpse at the potential of open medicine, and what can happen when families start to pair with scientists in the hunt for a cure.

Post has attachment
ABSTRACT (Mark Miller et. al)

Programs do good things, but also do bad,
making software security more than a fad.
The authority of programs, we do need to tame.
But bad things still happen. Who do we blame?

From the very beginnings of access control:
Should we be safe by construction,
or should we patrol?
Horton shows how, in an elegant way,
we can simply do both, and so save the day.
                                   with apologies to Dr. Seuss

Can anyone justify why we still use "pages" in the bibtex entries? Here is an example:

G. Altekar and I. Stoica. ODR: Output-deterministic
replay for multicore debugging. In SOSP, pages 193–
206, Oct. 2009.

Seems like an unnecessary detail. We all just google for papers, so what's the point? On the other hand, without pages I can save a line. October is also not needed.

Really, you just need: (1) title to get what the publication is about; (2) year to judge the relevance; (3) venue to judge credibility, i.e., the chances that the publication will have an important contribution, not just fluff, and of course to simplify search (especially for old work), and to distinguish blog posts, patents, white papers and peer-reviewed work; (4) authors, ideally the full list, not just the first author et al., to connect the citation to the previous work by the same people, and sometimes judge credibility as well. This is it. Four things. Everything else should go away. 

I think we need to make available a couple of stats which are not really discussed in the popular rants on "whether it makes sense to do a PhD".

First, I want to have  an easy access to an up-to-date ratio of number of people which apply for a single job opening in different areas at graduation level. Here is a couple of scary facts, which surprised me recently. Today in the area of cell biology, it's not uncommon to have 150 candidates for a single PhD-level opening in industry (some people say it can reach 400, but I don't have reliable data for that). Doing a PhD in biology is effectively a suicide. It's different for CS, so we can be calm for now, but a prediction for what will happen in 5-7 years can be helpful. There is some data from NSF, like this (, but it's hard to read

Second, I want to see a graph which shows the possibility of transitioning across different levels and across different institutions. For example, is it possible to get a faculty position in Utah after graduating from Rochester? Or really you need another hop like Rochester to MSR, MIT, or Stanford, and only then Utah. I mean, maybe there is no real chance of getting a faculty job after graduating from Utah. The access to such graph could make your PhD plans much more practical, e.g. if you got stuck in your school, quit right away with MS and try to do a PhD at a better school.

Post has attachment
Not your usual thermal paste failure. The thing on the picture is a boiler from Saeco Aroma. Lack of thermal paste screwed it up. One of the contact thermostats got loose, didn't measure the temperature of the chamber, didn't turn off the boiler, and destroyed the gaskets. It was fun to disassemble and analyze the failure, e.g. relate it back to how the machine behaved before failing completely.

Post has shared content
New well-deserved recognition for Professor Ivan Sutherland's pioneering contributions to the computer graphics technology that has become ubiquitous in the modern world. (Via +Michael Scroggins on fb.)

This reads as a John Steinbeck's book

Post has attachment
seL4 was acquired by General Dynamics, I guess it's time to start building an open source verified hypervisor
Wait while more posts are being loaded