spartango

Currently we don't have that much to show publicly as we're working primarily with doctors; we need to validate our technology properly before it gets anywhere near patients or the public eye.

With that said, if you shoot me an email (see profile) I'm happy to answer questions and perhaps more.

I'm so sorry to hear it. This is why we're working in cancer.

We all work as fast as we can, but I've already seen too many cases with poor prognoses. The hope is that in growing this technology, we can catch cancers earlier when they are more treatable.

I'm working on software that assists in cancer diagnoses.

We apply a healthy serving of computer vision and a touch of machine learning to high resolution images (10Gpx) of cancer tissue.

Hopefully we'll be able to make diagnosis faster and more accurate, doing our small part to save lives.

> Safety is very important to me. In medical diagnostic imaging it’s often important to have an image as soon as possible, but that’s not the case for scanning non-living objects purely for scientific or educational interest...[I] make use of a “check source”, which is an extremely low intensity source used to verify the functionality of a high-energy particle detector.

Thank you for doing this. It's easy to build something cool but extremely unsafe in your home, forgetting that medical devices are carefully controlled to ensure patient safety. All it takes is one irresponsible DIYer to cause problems for everyone.

Anandtech has a nice analysis of the progression of CPUs over the last 9 years: http://www.anandtech.com/show/7003/the-haswell-review-intel-...

Depending on what you're doing with CPUs there have been improvements, even at the high end between Nehalem and Haswell (on the Intel side). Anandtech's data shows a nearly 2x improvement between those generations for certain tasks.

But the reality is that for other tasks (e.g. the winrar compression benchmark) the speedup is minimal compared to previous years (see Haswell vs Prescott). In many cases where there is a speedup, performance is already so good that you wouldn't notice.

It's pretty crazy to see that in the consumer world, not even games are demanding so much performance from our CPUs.

Thank you for posting this.

Yes, it's not fashionable to follow the updated HHS guidelines for healthy eating, but they are rooted in well-tested science. The USDA and HHS invest quite a bit into getting these guidelines right continuing to improve them as science advances. I have not seen a dietary platform as rigorously studied as the one provided by HHS.

While my inclination is to agree with your critical view of social science research, I think picking at this example misses the point of the article.

The idea that 'hivemind' discussions are drowning out 'other' viewpoints has merit with respect to this forum. There are copious examples, from role-of-government discussions to technical preferences.

Indeed, there are cases where the community massively promotes dubious science that it finds agreeable.

We would be foolish to throw out this author's commentary because one of his examples is not bulletproof.

Good point, I forgot about this.

I should know this actually, I built OS X Live DVDs back in the day...

JRE 6 is the last JRE that Apple distributes itself via Software Update. JRE 7 comes direct from Oracle for the time being.

Unfortunately, this is not possible; 10.4 was the first release of OSX that supported Intel, and 10.5 was the last to support PowerPC. Alas, you can't have a machine that was on 10.1 that can also run 10.8..

That said, you can definitely chain update your way from 10.4 to 10.8 without issue, and similarly from 10.1 to 10.5. The latter case is a little tricky as many of the machines that ran 10.1 were not deemed fit for 10.5 (e.g. the G4 Cube variants), but there are workarounds.

Upgrading generally works great with OS X; generally the upgrade process blows away the system/OS files and leaves your stuff untouched. In the normal case, "your stuff" is well compartmentalized away from system files so stuff works nicely.

With that said, as a developer sometimes it can be handy to do a clean install. For example, if you have any custom kernel modules, these will certainly be blown away in an upgrade. Additionally, developer tools installed in /usr/ can be interfered with, and in general the probability that something will be incompatible/broken is a bit higher.

Potentially, but I've found the 'suburbs' of San Francisco/the Bay Area are very different from suburbs in other parts of the country. Suburbs are known for being somewhat isolated, uniform, and idle; the textbook definitions of suburbs describe them as a dissociation of residential life from industrial zones.

This is often not true around SF. Up and down the peninsula, you'll find continuous cities with strong tech and other industries. Would we consider Redwood City a stereotypical suburb, for example? It becomes more murky, and many of the stereotypes fall apart.

That said, your perception is held by many people and is accurate for many parts of the US. As a result, people moving to the Bay Area choose where to live with a similar bias.

While you're right about office space becoming wildly expensive in Palo Alto and Mountain View, I'm unconvinced the data supports the idea that cheap office space is the driver of growth to the city.

As you mention there is plenty of cheap and large offices as you go south along the peninsula, but startups seem to consistently choose more expensive SOMA offices over these. Only startups that need a warehouse or other specialized facilities start their search in the south bay.

I think the reality is that its a combination of factors driving people to SF, from 'cool factor'/perception to quality of life issues (employee proximity) to 'cheaper-than-palo-alto' benchmarking.

By the way, if you're reading this thread and looking for office space, the south bay is a perfectly good place to look. Yes, it's not as glamorous, but it's both functional, frugal, and actually quite nice. The graveyard doesn't consist of dumpy factories, but pleasantly maintained offices with good facilities. You can even live nearby, as long as calm neighborhood is an acceptable condition for you.

I wonder if what you're describing is that broad, or perhaps a slightly different phenomenon:

A huge amount of the growth of SF's (the city proper) tech scene seems to be driven by companies staffed primarily by very young (straight out of college, <30yrs) people. Similarly, much of the migration to the city seems to be driven by young folks (under 30/without family).

Back in the old days (when I was growing up on the peninsula), one of the big reasons people wanted to live in Palo Alto/Mountain View/Cupertino was that they were family-friendly towns with good schools and communities. Companies sprung up on the peninsula for many reasons, but people at or near child-rearing ages (~30+) had their priorities set.

I'm unconvinced that this rationale has changed in the intervening years. Gentrification and high demand have made living in the 'ideal' parts of the peninsula more difficult, but people still choose Palo Alto to start a family. Similarly, tech companies that tend to have slightly older workforces still prize their peninsula locations.

Perhaps I'm wrong, but I think studying the demographics of SF growth may tell us more. That said, I don't think that the suburbs[1] are being marginalized at all.

--

[1] While cities on the peninsula might be suburban in a way, they are quite different from suburbs in the rest of the country. It's a bit strange calling Mountain View a suburb when it is so different from a city like Needham (suburb of Boston) or the various suburbs of Seattle or Houston for example.

I would throw this question back at the OP (and the everyone else) as "for what?"

It's quite clear that languages are suited to certain tasks over others, which makes it nigh impossible to blanket characterize a language as 'good'.

Sounds like a management or business consulting arrangement, except longer-term and compensated with equity.

Management consulting is a fairly large industry, but there's an open debate about how effective it is, particularly for startups.

That didn't work so well for the Nazis; between the many track and field victories of the American Jesse Owens and excellent sportsmanship of many other international athletes, Nazi supremacy was at best marginalized.

Speaking of transcending political intentions, there's a story from those olympics where German long jumper Luz Long gave Jesse Owens tips to help him qualify for the event. There are countless other gestures that demonstrate the same spirit at that event.

To be fair, most of PBS' funding does not come from the government, but instead by donations from members (via local stations). With that said, I'm glad that there are still grants going to support PBS, as their work is excellent.

While an awesome company, Palantir is hardly the biggest or most prominent tech company in Palo Alto. Even though giant Sun Microsystems is gone, HP, VMWare, SAP, Lockheed and other giants remain. Loral Space Systems (by 101) is one of the city's biggest employers.

Indeed, as a general point, I do hope that Palo Alto is not known solely for Palantir, but also for its many long-standing institutions. From Stanford and the Palo Alto Unified School District's amazing (top-ranked) public schools, to its 6 libraries (for 66,000 people).

--A Palo Alto native (born and raised).

I grew up in the valley, and I can tell you that in 2007 the traffic on 280 and 101 going north in the morning was much lighter than it is today. As SF tech companies have grown dramatically in the last few years, so has the traffic. The same was true building up to 2001; after dotcom, the traffic went down dramatically. Of course, improvements to the highways and interchanges around that time made traffic feel even more sparse on the peninsula.

Those post-dotcom days were sad; the reason there was less traffic was the unemployment of people you knew. You felt that everywhere you went.

PiCloud is a layer of abstraction above the IaaS providers you've listed. That is to say that it gives you software infrastructure to design workflows and distribute them across the machines provided by an IaaS provider. In some ways it's like Heroku, and in others it's another layer beyond Heroku (App Engine-like?). You architect your work around the PiCloud tools. I think PiCloud currently runs on AWS.

I do not think it makes sense for the Fellowship to supplant research at big universities, or even compete with it. Flawless these are not, but there is much good derived from academic, basic science.

I do think that the Fellowship can be an interesting venue for young scientists to explore the cracks between basic science and industrial progress. A grounded student with a clever project can make good use of the fellowship understanding what it takes to bring that project closer to the real world.

Translational science can seem rather opaque, and being thrust into that world without a ramp to learn is hard.

2012 Thiel fellow working in science here.

I definitely don't see the fellowship as a substitute for a proper academic training in the sciences. A solid training is important; I have benefited greatly from my academic roots and support (Harvard).

In that context, the fellowship represents a supplement to my academic training, exploring the translational science that lies between basic research and commercialization.

I'm taking a detour from the traditional course of study primarily so that I can spend some of my time working in the aspects of translation that extend beyond the lab. In my biomedical field, that is everything from medical workflow to regulatory affairs to product pipelines to validation studies.

More broadly, I'm doing this exploration in the context of a tractable project that has involved both research and commercialization. It's at the intersection of my scientific interests, and is fueled by a desire to help people in a tangible way. The timing is such that it makes sense for me to dive into it full-time with my existing training.

I'm still learning many things, some in academic research settings and others in commercial settings. The fellowship simply serves to focus my limited time to the intersection of those two areas.

You bet I'll be back in academia after this, but in the meantime, taking a small bite out of cancer seems a worthy detour.

Feel free to shoot me a mail (email in profile) if you're curious.

The article cites Cloud Market's data on AMIs, which indicate how many AMIs have been derived from the Ubuntu base AMIs, rather than actual deployment numbers. It's an interesting indicator...

Having been on the inside (EC2) and seen the numbers, I can confidently say that Ubuntu's AMIs are the most popular AMIs deployed on EC2. The stock AMIS take the top spots by an overwhelming majority, and most of the lower spots are filled with derivatives of the original Ubuntu AMIs.

Ubuntu's been working hard in the server/cloud space for a while, between the JeOS effort and subsequent, excellent tools for building customized installations around it. It's not just a brand or that makes them popular; with Ubuntu you get a sane system, style, and toolset for your system images.

It really depends on what you value to be "worth it".

If you value the end goal of becoming a professor, you'd agree the Phd is "worth it"; perhaps this provides a career-oriented perspective.

But consider the other values:

If you want to explore a novel, foundational (perhaps theoretical) area of the field, academia represents a good venue. The journey is what is "worth it" to the value of intellectual curiosity.

Similarly, if you value gaining extreme depth in a narrow topic area to the end of advancing that topic area, academia enables an appropriate level of focus for that.

Blanket statements of what academia is useful for are biased by individual values, and are thus generally unhelpful. Not everyone pursues a path for its ends or career utility.

Yea, they eschew the DRM-only TPM part to store the keys in the SMC, which does a whole host of things and doesn't protect the keys. I suppose you were sufficiently vague ;)

There's just an outstanding myth that Macs still use TPM...

OSX does have 'protected binaries', which refers to binaries that use encrypted pages[1] that require decryption via an anti-piracy kernel extension (Don't Steal Mac OS.kext). This extension can check that you're running on Apple hardware by checking with the System Management Controller(SMC), which is Mac-specific[2].

In the old days (10.4), hackintosh systems would patch the protected binaries to remove the page encryption/decryption. Over time kernel extensions were developed to replace the functionality of Don't Steal Mac OS.kext (dsmos.kext or AppleDecrypt.kext) without the hardware checks.

The reason you no longer need either of these extensions is because a kernel extension called FakeSMC.kext was developed, which emulates as much of the functionality of the SMC as possible. This includes thermal and fan monitoring as well as the decryption key storage.

In a modern hackintosh setup, you make use of EFI emulation (inc. DSDT support) included in the bootloader (a boot-132 derivative, most likely Chameleon) and the pre-boot loading of emulator kernel extensions like FakeSMC.kext.

[1]: http://www.osxbook.com/book/bonus/chapter7/binaryprotection/ [2]: http://osxbook.com/book/bonus/chapter7/tpmdrmmyth/

The TPM was used briefly at the beginning of the Intel mac days, but hasn't been used since. The decryption is now done through dsmos.kext in software.

http://osxbook.com/book/bonus/chapter10/tpm/

This used to be the case, but no longer is; Apple stopped using the TPM in Intel macs shortly after they launched[1] and have not brought it back since.

DSMOS.kext has another function, however; it is involved in Apple's binary protection scheme[2] for certain OSX binaries. Essentially, OSX ships with certain important binaries (such as loginwindow, SystemUIServer, Finder, and Dock) that have certain vm pages encrypted. Decrypting these pages for use requires DSMOS.kext.

[1]: http://osxbook.com/book/bonus/chapter10/tpm/ [2]: http://www.osxbook.com/book/bonus/chapter7/binaryprotection/

Note that Quest Diagnostics has a drugstore partnership with CVS[1], directly competing with Theranos+Wallgreens. Labcorp has similar offerings, although some are whitelabeled.

[1] http://ir.questdiagnostics.com/phoenix.zhtml?c=82068&p=irol-...