[Beowulf] Off topic - Feynman and The Challenger
j.sassmannshausen at ucl.ac.uk
Thu Feb 13 05:10:09 PST 2014
as a computational chemist (and sys-admin):
I agree with Jim's last paragraph here (and the rest of the email). We
computational chemists don't really talk about the usage of HPCs for number
crunching. However, what I am doing is I am giving tours in the data centres I
got, tell people in simple terms what the machines are doing ('we investigate
catalysts which make carrier bags and we are currently working out how to make
them with BRANDNAME HERE is printed on it') and all of the students who have
seen the machines were impressed. I think that is a bit of the problem we are
facing: for good reasons these clusters are locked away so people don't see
them. Seeing is believing. I tell people in plain English about the difference
between ethernet and IB and make a joke about: that 'this my high-speed
connection between the two racks (pointing to a large flexible tube) which we
got in place unlike HS2 (you need to know that HS2 is the proposed and
controversial High Speed rail link here in the UK)'.
What I try to say is: we need to get the public engaged into science so they
know why we need a new computer system (in my case it is old hardware and a
huge energy bill) and what the benefit would be for them (a warm staircase due
to heat recycling).
My two cents here
On Thursday 13 Feb 2014 01:33:35 Lux, Jim (337C) wrote:
> On 2/12/14 11:51 AM, "Douglas Eadline" <deadline at eadline.org> wrote:
> >FYI, I liked it, found some things a little forced to
> >squeeze in some references to Feynman's other work.
> >I still recall seeing him in the televised hearings
> >demonstrating the frozen sealant and thinking that
> >is such a brilliant and simple demonstration of
> >the issue.
> A trap that technologists (including myself) fall into is getting
> fascinated by the technology when explaining something to non-specialists.
> We go on about sophisticated variable rate coding schemes that allow
> approaching the Shannon Limit with minimum computational resources and
> iterative maximum likelihood decoders with parallel processing, and what
> they want to hear is "HD video on your cellphone"
> Feynman was very good at working at multiple levels of explanation, and
> that is not easy. So while Feynman could probably hold his own talking to
> a bunch of mechanical engineers talking about elastomer nonlinear behavior
> over temperature, he knew that a simple demonstration of that behavior was
> more appropriate for that audience.
> He did get flak for being a bit of a showman for that "stunt", but I think
> that's the same thing that occurs for "science popularizers" like Asimov
> or Sagan. It's almost like a hold over of the "amateur tradition" from
> Victorian times: Gentlemen do not seek attention, it is vulgar.
> I've had the opportunity over the last few months to spend a lot of time
> explaining a moderately complex technology to an interested but
> non-technical audience. As a result, I've got a heck of a lot more respect
> for people who can do it well. (google/bing "jim lux NASA heartbeat")
> Interesting to this group, we DID use HPC in developing FINDER: FDTD EM
> propagation models. However,I don't talk about Rayleigh scattering and
> dispersive heterogenous media and model/experiment correlation. I have a
> physical model that I can hold in my hand to show what's going on, albeit
> somewhat inaccurate in scale.
> I think it's important that if we want people to spend money on high
> performance computing and other nifty technology, we've got to make it
> tangible to them in a desirable way. Sadly, many of my acquaintances
> think that HPC is mostly used for high frequency trading (evil bankers in
> the 0.001%) and (recently) Automatic scanning of phone call and email
> content (evil government snooping), and don't think in terms of "better
> fuel economy for your car". Computational chemistry and biology are also
> big users of HPC, but I avoid talking about those, because that can
> trigger the whole genetic engineering/big Pharma discussion.
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit
University College London
Department of Chemistry
email: j.sassmannshausen at ucl.ac.uk
Please avoid sending me Word or PowerPoint attachments.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 198 bytes
Desc: This is a digitally signed message part.
More information about the Beowulf