Hacker News new | past | comments | ask | show | jobs | submit login
John Hunter [matplotlib.sourceforge.net] has died. (scipy.org)
324 points by archgoon on Aug 30, 2012 | hide | past | favorite | 32 comments

Numfocus is collecting donations in his memory for the education of his three daughters: http://numfocus.org/johnhunter/

My startup uses matplotlib and I was quite affected reading the story of John Hunter's last days and hours. (It also got my attention as I happen to have an elevated colon cancer risk.)

So a small donation is a no-brainer [and done] but I want to do something more. We're only recently ramen-profitable so I'm thinking it would be nice to donate some shares. Beeminder's success will continue to be thanks in part to John Hunter's work (even if we port to d3.js which I confess is in our plans) so I like the idea of having my donation proportional to the magnitude of that success.

I suspect it's difficult to formally donate shares so I'm actually writing this comment as a commitment device to honor the following:

I, Daniel Reeves, hereby earmark 0.1% of my own Beeminder equity for the John Hunter Memorial Fund, to be converted and paid as cash within 5 years of today, 2012-08-30.

Also, anyone with colon cancer in their family should please do a colonoscopy every 10 years from the age of 40 to greatly reduce the likelihood of not detecting colon cancer early. Early detection has a recovery rate of near 100%.

Wow, ONLY every ten years if they're at high risk? I'd think that you'd want to do it at least every five years, if not every year.

Colon cancer typically grows slowly. If you are at risk, and get a colonoscopy at 40 and they find polyps then they will have you come back to be sure they don't come back.

Basically at risk people should get checked early and you're at risk if you've had an immediate family member have polyps or cancer.

My grandfather died of colon cancer and I know my Dad gets a colonoscopy every single year, and I suspect I will too once I'm a bit older.

Depends on the risk of the person - some need it every 5 (or more), and some much earlier than 40. But generally speaking, it takes ~10 years (average) for an initial growth to become "fully" cancerous. Hence the 10 year rec.

Still - point is this is one of the few forms of cancer that can be treated extremely (simple exicision) easily with early detection. Its common and (once advanced) deadly.

Good on you!

I used matplotlib extensively when I was doing research back in school. It even made it into the work I published (after I got ridiculed for using excel to make plots). Being able to pre-process my data with numpy and generate all of my plots automatically via matplotlib calls kept me sane.

John's contribution to the Python community was extensive (I don't know of any comparable plotting tools) and won't be forgotten.

John Hunter was, like me, a neuroscientist on paper. There were times when he helped me figure out how to best plot spike rasters (neural data) in matplotlib. I remember him leaving academia for industry (quant finance at the time), and discussing how he really needed to be able to provide more for his family. Those discussions ultimately stuck with me, and now having left academia myself, I often looked back on them as catalyzing. Thank you, John.

He recently gave a talk reflecting on matplotlib: "Lessons from middle age"


Edit: At the end, he discusses the challenge of client-side rendering to make visualizations like the New York Times.

d3.js comes to mind, a javascript library with a structured approach to scales, axes, projections, layouts and data-binding with SVG. The project has 12 years of work in it according to ohloh:

http://d3js.org/ http://www.ohloh.net/p/d3js

I wonder what middle age will look like for javascript communities.

A spiral for John: http://bl.ocks.org/3543186

I think today it is quite understated the value of giving back to the programming community. It is so easy for people to consume and take the information online but it is just as important that people contribute as much as they can back to community to continue the cycle. John is a perfect example of someone who went beyond the call.

I had the pleasure to work with John some a while ago when I was part of the Sage Math project.

Very saddened to hear this news :(

(Just donated. I'd urge anyone who has benefited from John's amazing Matplotlib project to considering donating to help his family).

I found matplotlib to be extremely helpful in my research...I also just donated.

I see lots of articles like this but none have ever hit me like this one. matplotlib saved my ass so many times and there is no doubt in my mind it was a huge factor in my career success.

Having a daughter in December and all I want to do now is call my doctor for a checkup.

Thank you John.

Very sad to hear. His contributions to the Python community will never be forgotten.

As Fernando noted, if you use matplotlib regularly, check the price on a MATLAB license (looks like USD $2150.00 for a single commercial user) and then consider donating to John's family: http://numfocus.org/johnhunter/

I use matplotlib almost every day for work and school. I'm very sad to hear of his passing.

Donating to help. His work helps me so much, this is the least I can do.

matplotlib got me through my physics undergrad degree.

I didn't know about this man until today but I still feel a loss.

very sad - I will be donating. I would not be doing what I am doing today without matplotlib - a scientific computing toolchain without a good visualization component is a non-starter.

Wow very sad news, RIP. I never knew him personally but feel like he gave me so much through his contributions to the python and scientific computing community.

Did the hospital use a medical checklist? This is a question that I like to ask everytime a medical procedure went wrong, because a checklist is a very useful tool in saving lives and preventing medical errors. However, the medical checklist haven't been adopted as fast as it should.

I don't think it is helpful in this situation to insinuate that medical error was involved, or that Mr Hunter's death was preventable. Also, thinking that a simple checklist will solve medical error reflects a very simplified view of the problems in the health care system. There are a lot of smart people in health, if the solution was simple the problem would have been solved by now.

Not knowing anything beyond the post about Mr Hunter's situation, in general advanced colon cancer may involve surgery (to prevent blockage) and chemo. These are risky and have high chances of complications. In this sad situation there are no good options, only less bad ones. From the post it appears he fought bravely and was dignified. Very sad.

> I don't think it is helpful in this situation to insinuate that medical error was involved, or that Mr Hunter's death was preventable.

I agree.

> Also, thinking that a simple checklist will solve medical error reflects a very simplified view of the problems in the health care system.

Indeed, yet, a simple checklist HAS been shown to significantly improve outcomes: http://www.npr.org/templates/story/story.php?storyId=1222261... .

> There are a lot of smart people in health, if the solution was simple the problem would have been solved by now.

This is completely and utterly wrong: All stable systems past their transient stage are generally at a local optimum for their respectively relevant utility function - which is shaped by the incentive structure.

Unfortunately, the incentive structure in medicine is all about avoiding liability, then making money, and actual care is a side effect. As a result, small changes can produce dramatic change in care quality. But (almost by definition) they increase liability in the short term regardless of merits (because "that's not standard practice"), and are rarely practiced.

I've spent the last 20 years in the area of patient safety and reducing medical error -- the last 10 years in building technology to assist clinical staff in improving safety. You are correct in saying the incentives are not right, but regardless, many people are in health to help people. Reducing error is hard because the 'system' isn't designed for safety (it basically isn't designed).

Checklists are used often, but aren't the solution. There are about 100 things that have to line up to get health right, saying 'checklists _alone_ will solve it' trivialising the problem.

The challenge is how to improve patient safety while improving clinical workflows. Everyone is under a lot of pressure and adding more stuff for them to do just fails.

While i agree that there's nothing to indicate there was neglect or that this death was preventable, please please do not assume that checklists would not help. Doctors and nurses are humans (usually exhausted over worked humans). Checklists save lives and make sure that folks are all on the same (literal) page in aviation, spaceflight, engineering fields of all stripes and other fields where life and limb are at risk.

Just because you're smart doesn't mean you can't make mistakes. Engineer your environment to minimize mistakes.

As tragic a loss as this is, I'm not sure that this was a simple 'skipped check' complication. Chemotherapy does sometimes result in complications like organ failure even if no actual errors are made.

I'm not a physician, but I'm did fly as a pilot and learnt all about checklists. Capt. Sullenburger would like to apply the lessons in his 40 years of flying, to the healthcare industry.

Medical professionals are often dragging their heels on progress as a matter of course, but in this case I believe the allegation is unfounded. Chemotherapy is a very dangerous treatment. It's essentially an injection of pure poison, and often people do die from complications of cancer treatment as opposed to dying from the cancer itself. Cancer drugs are terrible. Cancer surgery is often devastating. The worst thing (though maybe somewhat of a consolation in this case) is there is no real cure.

If he was diagnosed with advanced colon cancer and pushed straight into chemo, chances are the treatment was just a palliative measure designed to give him a few more weeks with less discomfort.



Such a shock to hear this. I was lucky enough to have worked with John at TL. He was a generous and kind man, always happy to help a novice programmer advance his or her Python skills. He will be missed. My sincere condolences to his loved ones.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact