HNHacker News
TopNewBestAskShowJobs

ephermata

221 karma · joined June 27, 2009

David Molnar

http://dmolnar.com

I previously posted under the account "dmolnar" as well, but this is my main account.

Please consider donating bone marrow. You could help someone beat leukemia. Learn more & join here: http://join.marrow.org/

submissionscomments
ephermata··on Apple's Steve Wozniak: 'We've lost a lot of control'
I have multiple phones on me at any given time. If you want to do side by side comparisons of how the phones "feel," or if you want to try applications that only exist on one phone platform, then you need to do this. eBay is a good resource here, because you can buy phones without contracts at reasonable prices, especially phones that are in the immediate prior generation.
ephermata··on Matt Cutts: $1,000 to the 2 best implementations of the recently-hacked Kinect
The new XBox S (the matte black one) is much quieter than the old XBoxes. This YouTube video gives some idea of the difference, throwing in a listen with a PS3 as well: http://www.youtube.com/watch?v=K5QpoKwVcSk

The Kinect experience so far is targeted at a wider audience than the traditional gamer. My mother, for example, absolutely loves the Kinect. She particularly likes the Tai Chi and yoga trainers in the "Your Shape" game. I didn't expect it to take off with her the way it did, but it's the first video game system where she really wants to play it.

Another thing: there are several different interfaces embodied in the launch titles around the same basic Kinect capabilities and interactions. If you want to develop new user experiences, now is a good time to check out what people are doing before everything settles into place. So it's worth at least visiting a friend who has one.

(Disclosure: I work at Microsoft. Speaking only for myself here, of course.)

ephermata··on Android
Not quite what you are getting at, but a lot of university-led CS research uses Android. Android hits a sweet spot where it's easy to hack, but at the same time changes you make are relevant to millions of shipping devices. Security research in particular has a lot of fun with Android, because you can analyse the existing platform, then turn around and build your idea, then turn around and run it on real hardware without too much pain.

That does not, of course, necessarily filter into intro CS classes. I have seen a few "mobile development" classes in universities, but they seem to be split between iPhone and Android.

ephermata··on Soylent: Mechanical Turk for Microsoft Word
This is wacky beautiful. I look forward to the packaged precompiled plugin so I can try it out myself.

One thing that comes to mind -- it may not be that hard to overwhelm the Mechanical Turk tasks associated with Soylent to reach a situation where I see every single Soylent task from a single document. It's an interesting question to calculate how many people you would need, given the volume of Soylent usage.

The point is that if I control every single worker on your Soylent task, then I can lean a lot about your document. That in turn could lead to exciting conversations with your IT department.

ephermata··on Inflation in job titles reaching Weimar levels
My friend and I ran a chess club in high school. We really just wanted to play chess, but made everyone who joined a "Director." Why not? Plus, It was a bit of a joke reaction to the high pressure competition we saw in other extra curricular activities.
ephermata··on PhD or Not PhD: that is the question
Stanford CS PhD has few or no course requirements, or at least they did back in 2002.
ephermata··on Jason Calacanis' Warning To Y Combinator Startups [video]
Maps are an interesting example because map data is actually hard to replicate. Google went ahead and built their own database anyway, after several years of licensing from NavTeq and then Tele Atlas. http://blog.daveburrows.com/2009/10/08/google-drops-teleatla...

This allows Google to do whatever they like with the data, including offering turn by turn directions on Android.

ephermata··on How to Screw Up an On-Site Interview
My first interview for a technical job, I showed up in a suit and tie. The rest of the interview had each person start by asking some variant of "wait, are you sure you're interviewing for engineering and not marketing?" The feeling this conveyed was that if you wore a suit, you couldn't possibly be a real engineer.

I landed the job - a summer internship - but the suit was a distraction. It certainly did not help me relax and perform my best. The job rocked, so I'm glad the people there were able to look past the suit. Since then, I've taken to asking people what kind of dress code they expect on an interview.

ephermata··on Day in the life of a Galois intern
Thanks for the great day in the life post. Would be interested to see a similar post focusing on the technical aspects of the project. Maybe that's the tech talk, but a post would be helpful to for people who can't make it in person. : )
ephermata··on Google web security expert posits the "Millenium Prizes" in web scanning
Some of these problems would be aided by instrumenting the web application itself. Of course instrumenting a legacy web app is its own set of hairy problems, especially if you want to handle production servers, but even a little bit of instrumentation and logging might catch exceptions or map requests to methods actually executed.
ephermata··on Microsoft Street Slide: Street view will never be the same
Whatever you want to use for the project. For a recent project, one of my coworkers created a domain specific language and the rest of us coded in that language. For another project, I use C++, because I'm leveraging an existing code base. Colleagues of mine use F# for program analysis tools, which they love. As you say, F# came out of the MSR Cambridge languages group, that's a great tech transfer story.

That being said, there are a lot of internal libraries or projects written in C/C++ or C#. So I see a lot of those languages. I've also been learning ASP.NET, in part so I can better understand cross site scripting attacks on applications written on that platform.

ephermata··on Microsoft Street Slide: Street view will never be the same
Fortunately, that assumption is false: junior researchers can start their own projects. I joined Microsoft Research a year ago. In that time, I've been encouraged to come up with new projects where I can lead, as well as make contributions to existing projects. As you point out, this degree of independence is rare in industry, so it makes MSR a competitive place to work.
ephermata··on PiCloud Launches Serverless Computing Platform To The Public
Congrats to the PiCloud team - looking forward to more.
ephermata··on HTC 1
That image reminds me of the Zune HD interface, actually. Same "wall of text" approach. I've noticed it's a love-it-or-hate-it approach. I happen to like it.
ephermata··on Google Secretly Invested In Zynga, Preparing To Launch Google Games
Is it possible to buy 7-11 coffee with points from zynga games?
ephermata··on How Microsoft Research banished the BSOD
In addition to Prefast, there is also the Static Driver Verifier(SDV). Comes with the Windows Driver Kit, analyzes your code to detect common problems people have when writing drivers. The Windows Driver Kit is free -- you can learn more about SDV here: http://www.microsoft.com/whdc/devtools/tools/sdv.mspx

SDV began its life as a project called SLAM. This page has documentation on SLAM, academic papers, etc. http://research.microsoft.com/en-us/projects/slam/

I'm not sure of the exact connection highlighted in the article, but this work builds on decades of work in automated theorem proving. It's exciting that now these techniques are used on real code.

(Disclosure: the SLAM & SDV authors are colleagues of mine at Microsoft.)

ephermata··on Somebody please hire me before I take a job in weapon development
Speaking for the U.S. context only: once you have a security clearance, which will become necessary at some point for defence work, it becomes difficult to move out to non-classified work. On one hand, the pool of people with clearances is smaller than the general labor pool, so you can have a higher salary & benefits, therefore it's more attractive to stay at that kind of work. On the other hand, prospective employers outside defence likely do not have clearances and so cannot learn anything about what you did - this makes it difficult to leverage your experience when seeking to move to a new job.

I don't work in defence. That is just what I've heard from people who do as one of the reasons why it's hard to "rotate in, rotate out."

ephermata··on Google's getting B(l)ing
My mother regularly tells me how much she loves the Bing photos and how she's looking forward to "what will they put up next." She was asking yesterday how long it would be before Google did the same. Now we know...
ephermata··on List of Hacker Spaces Around the World
Somehow I missed this - thanks much for putting it together!

Noisebridge is one of the things I miss about the Bay Area.

ephermata··on The Secret Lives of Professors (What Being a Professor is Really Like)
Yes, that's right, the writing up is key. In addition to spotting holes in arguments, it's good for showing you that you can finish a large sustained project.
ephermata··on The Secret Lives of Professors (What Being a Professor is Really Like)
What settings have you found where you could call yourself "doctor" without someone mistakenly believing you are an MD? The only ones I've found so far are in academia and at work, but in both cases it's not really that much of a boost. (At work everyone more or less has a PhD.)
ephermata··on The Secret Lives of Professors (What Being a Professor is Really Like)
A PhD programme is ideally an opportunity to work on whatever you want, for years, in the company of smart people who are also interested in the same subject. You can use this time to try out different ideas and see if any of them take off. Having that shot is valuable in itself, even if you don't "succeed" with a research position afterwards.

Plus, sometimes new startup ideas, connections, or key skills come out of a PhD. The traditional approach is to found a startup directly from your research, but equally important is the role of PhD projects in helping you identify people who will get things done. If they flake out on helping write an academic paper, what will they do when you're working on a company?

Now sometimes a programme falls short of the ideal. You may end up teaching too much, you may end up working on projects you don't like because the funding is there, and so on. If that's happening, then that is a reason to quit. If you have an incredible opportunity that comes up and isn't compatible with continuing, that's a reason to quit. Just quitting because the grass is greener, though, does not seem like it will work.

ephermata··on I am a junior angel in Silicon Valley. AMA
What are the problems with bad entrepreneurs that keep you from seeing the pitch itself?

Thanks for your time answering these questions.

ephermata··on Code as photo-op
"A 2005 Times article by John Markoff about a new startup that aims to commercialize podcasting. Didn't happen."

The article is on Odeo. The people involved seem to have turned out all right. This is a joke?

ephermata··on Nowmov (YC W10): Sit Back, Relax, And Watch An Endless Stream of Videos
I upvoted it because I worked on a similar idea with abossy and randomwalker. I agree with abossy's comment above: this is a great idea, I wish them all the best and I hope they kick ass, but the business side is difficult. So long as you rely on YouTube for videos, you are at the mercy of changes to the YouTube terms of service. You can look for other sources of videos besides YouTube, but this is challenging.

So I'm hoping they succeed where I failed. Hence the upvote.

ephermata··on Rest in Peas: The Unrecognized Death of Speech Recognition
While the article focuses on speech recognition for arbitrary speech, it misses the fact that speech interfaces for specific situations are now actually useful. Today, I told my car "play track 'severed head'," and it actually played the correct song. I asked my phone "what is my next appointment?" and heard my calendar. I then said "dial 206 421 8989," my phone dialed properly, and so on and so forth. This is all without any explicit training on my part. No need to read Mark Twain or anything like that.

There are still problems here, but the technology for speech interfaces has gone from terrible to OK in the last seven years. I'm looking forward to seeing where it goes next.

ephermata··on Facebook is a Ponzi Scheme
My apologies for not being clear: Facebook ads is far from the only method in use for enlisting potential donors. Furthermore, families in need typically do not need to do anything because of the resources available. In my case I choose to do more because I am fortunate enough to be able to do so.

In the United States, the Be The Match Registry, operated by the National Marrow Donor Program, employs full time recruiters to encourage people to sign up for the registry. One of those recruiters, Magda Silva, has been tireless in her efforts to find new donors for my sister and others. For example she organizes marrow drives on at least a weekly basis where people can sign up to have their tissue type tested. Immediately after hearing about my sister, she worked to set up a drive a Noisebridge, a San Francisco hacker space.

In addition, transplant centers, such as the one at the Seattle Cancer Care Alliance where my sister is being treated, have resources to find potential donors. For example, because our family has significant Hungarian background, the transplant center is reaching out to the Hungarian national donor program. These searches are targeted for the patient in need and carried out by doctors who are intimately familiar with the search process.

These efforts are amazing and have paid off for thousands of people in need, including my sister. The Be The Match registry is a nonprofit, supported by generous donations. The transplant center is supported by health insurance and also by generous donations. Neither my sister nor I have been required to do anything special to find donors with these resources.

Indeed, after I began the ad campaign, the transplant center found a partial match for my sister. My family and I are so grateful that someone decided to voluntarily donate their marrow for a complete stranger. We are also grateful to the registry and our transplant center for all their work that made this possible. (What's more, someone even donated cord blood that matches her, which is another, more experimental alternative.) Thank you, whoever you are!

So why bother with online ads at all?

0) When I started, I didn't know if we had a match. I could do something by using these ads, and I was sure it would not duplicate existing efforts (see below). I am lucky enough that I can afford it.

1) Targeting for my sister's marrow type. The existing recruiting is by necessity broad-based, trying to bring in as many donors as possible from as broad a spectrum of people as possible. To the extent there is targeting, there is an emphasis on bringing in minorities because they are critically under represented in the registry. If you are African American and need a marrow transplant, you are likely in serious trouble.

In contrast, I can target ads to geographies which are most likely to have people matching my sister's tissue type. For example, there are many people of Hungarian descent in Ohio, our family has a heavy concentration of Hungarians, so I can target ads there.

2) Potential to scale up. Despite the best efforts of the registry and transplant centers since 1988, only 7MM people total are signed up in the U.S. registry. The in-person drives I mentioned above typically bring in between 30 to 200 people per drive. Roughly 12K sign up per month, principally through in person drives, with a smaller portion through the web form.

This means if you don't find a donor immediately, the chances that a donor will appear while you wait are small. The promise of online ads is that they could potentially scale up better than in-person drives -- with a large enough advertising campaign, we could acquire more donors per month than is feasible with current methods. How large a buy is needed of course depends on the conversion rate (i.e. signups for the repository) of ads. That conversion rate is one of the things I am learning through this experience.

I do not mean to suggest that ads could or should replace the traditional in-person methods. Certainly they will not. What they can do is offer an additional tool to attack this problem (and hey, maybe an effective way for some large advertising company to donate in-kind to a good cause).

Does that address your concerns? Again I'm sorry for not being clear that this is not the only method available to our family for finding donors. Don't hesitate to ask any follow up questions.

ephermata··on Facebook is a Ponzi Scheme
Yes, I have had people click on FB ads. It's not clear to me if they are more effective than alternative ads, but people are clicking on them. The ads are not traditional, though.

I have an ad campaign running now on Facebook encouraging people to go to http://join.marrow.org/gosusan . ("My sister has cancer. Help her and people like her beat it by donating your bone marrow! Click here to sign up!" )

Short answer is I've yielded 6 people signed up out of ads from Facebook, Myspace, and others. The signups could be improved, but it looks like at least one of these methods is recruiting donors.

Overall Facebook campaign since 9 March 2010:

4,801,855 impressions 803 clicks CTR 0.017% Average CPC $0.99 Spend $796.99

My banner ads on MySpace ("A CANCER KILLA IS YOU! My sister has cancer. Help her or someone like her kill it by donating bone marrow. Click this ad to get started!") overall have these numbers:

973,774 impressions 452 clicks CTR 0.05% Average CPC $0.45 Spend $203.00

At first blush MySpace looks like a better deal (compare the CTR). The thing is, for targeted geographies, I am seeing a lot more clicks, faster, from Facebook than from MySpace. That is, while the clicks from Myspace are cheaper, I'm seeing fewer of them.

My most recent campaign is targeted to south Ohio and south-east Ohio. (Turns out this is where a lot of people with similar heredity as me and my sister live. Therefore I expect these people to be more likely to have a matching bone marrow type). The Facebook ad for this geography has yielded 95 clicks in the last two days. The MySpace ad has yielded 10 clicks.

ephermata··on Ask HN: Would you use a cord bank?
Cord blood can today save the life of someone with leukemia or other diseases that require bone marrow transplants. While other cord blood stem cell therapies may be speculative, this one is real and already works to save lives. My sister has leukemia and we just found out that some kind person donated cord blood which matches her tissue type...I can't describe how grateful I feel now. Thank you, whoever you are.

In my case, for my children, I would certainly make sure the blood is saved. The only question is whether to do private cord blood banking or to donate the blood to the public blood bank. I would find out how rare is my child's human leukocyte antigen (HLA) tissue type as part of making the decision. If it's rare then it becomes much, much more important to bank known compatible blood.

One confounding factor: if your child turns out to have a blood disease later in life, it's not clear to me whether there is a risk in using your child's own stem cells. For example we know little about the causes of leukemia, it could be that even the cord blood cells are potentially cancerous. I'd suggest reading up on this and reaching out to researchers in the area on this front, since it's unlikely your OB/GYN would know the most recent results.

ephermata··on Don’t Hash Secrets
Where does Slide or RockYou host?
← PreviousPage 2 of 4Next →