<LarryHunter> Lesson 1: User Centered Design is important to this community, and I need to read Steve Sawyer

<LarryHunter> Lesson 2:New tools, IRC for collaborative work and Mendeley for bib management

Didn't know that the distinction between science and discovery was so important.

<LarryHunter> Lesson 3: NSF might actually fund my work, since while the applications are biomedical, the computer science is generalizable and may be of interest to them

I found the challenges in social science "discovery" extremely interesting... and would like to follow up on them!

<Kerstin> Strong emphasize on biology,in oposition to other meetings where physical sciences dominate

<Karsten> +1 on bio emphasis

<PaoloCiccarese> 1) Importance of usability and user centered design

<helenadeus> Lesson 1: formal knowledge representation is really important, but as a starting point, not an endpoint

<helenadeus> Lesson 2: Don't ask scientists what tools they need because they don't know

<Yolanda> 3 points from NIgham

<helenadeus> Lesson 3: Workflow can be used not only for integrating computational tasks, but human tasks as well

<Yolanda> 1) tradeoff between formal representations and the effort they take, can we quantify/formalize that

Demonstrating utilty is tricky

<Yolanda> 2) determining impact of tools is hard -- how do we measure impact

<Yolanda> 3) Need to work with scientist to design appropriate tools (Turbotax example)

<Yolanda> Kerstin points

<Yolanda> 1) diversity of backgrounds in meeting (I think that is what he said)

<HuanLiu_> @Huan 1 - Discovery has its many definitions and types

<Yolanda> 2) Scientists often perceive CS as not research

<Kerstin> where DOE program managers invited

<Yolanda> 3) had not realized the importance of knowledge representation (in climate)

<Karsten> 1) I did not realize there had been so many "success stories" in other science domains -- helps misperception that CS is 'just a service'

<Karsten> 2) Wasn't aware of the many different working definitions of "discovery" -- first hour felt like trying to define and scope

<Karsten> 3) Realized how little attention I have paid to knowledge representation -- we often just "make it work" for the task at hand

<Yolanda> Loren's 3 points:

<Yolanda> 1) many people did not know of success stories

<Yolanda> 2) -- I missed that

<Pat_> People were far less focused on induction from large data sets, and more balanced in their concerns, than I expected.

<Yolanda> 3) hard to work together when people have diverse backgrounds

<Loren> We all have unique knowledge.

<Yolanda> 3 pints from Evelyn

<Yolanda> 1) importance of discovery for so many people

<Yolanda> 2) importance of social computing

<LarryHunter> +1 to three pints

<HuanLiu_> @Huan 2 Social computing may be able to help discover global knowledge from local findings in a tangible way

<ChrisS_> +1 to pints (of beer)

<Loren> Learning: We all have unique knowledge.

<Loren> Learning: Social computing as a first class computing method.

<Yolanda> 3) contrast between industry and academia: a) in industry anyone can be an innovator

<Pat_> Analyses of the entire scientific process can serve as organizing frameworks for research on computational discovery.

<Loren> Learning: Renewed emphasis on working on meaningful issues.

<Loren> Learning: Difficulty of getting a group of people to work effectively and constructively together in a meeting.

<Kerstin> would have liked to define the boundaries of social computing applicability more clearly

<PhilBourne> Breadth of expertise

breadth of expertise... some essentially addressing same problem

know should be sharing data, have good data formats... but either quick and dirty or formalize

<HuanLiu_> @Huan 3 So much interest in social computing and so much expection of its power.

are there enough commonalities to formalize and build shareable tools across communities

<PhilBourne> sharing data either quick and dirty or fully formulated - makes sense to formalize sharing tools and then adapted to communities

<PhilBourne> above is from Alex

dilemma, fancy CS tools not being used. User-centric design approach might be the right thing to do.

<PhilBourne> Vipin

<PhilBourne> so many aspects of informatics that I did not know about

so many other aspects than DM/ML in informatics and how CS can help

got to know people in other areas coming from different aspects

commonlatiy of experiences in room

<PhilBourne> commonalityin experience s in different doamins

<Karsten> Alex Schliep: 1. breadth of expertise, but have the same type of problems 2. know we *should* use formalism, but usually do it quick and dirty 3. bioinformatics has lots of nice tools, but many of them dont' get used very much -- user-centered design could help this.

<PhilBourne> astronomy a good model for users vs experts many users per expert

need success stories to sell to colleagues in other disciplines... CS is not a service discipline

<DavidJensen> Incentives for participation really seems like an important issue. Researchers may not *want* to share workflows, data, and tools. They may see these as their competitive advantage.

<PhilBourne> Yan

<PhilBourne> computer science viewed as a service discipline must work to change that

started working closely with one biologist on single problem, not that many people cared about those results

<Karsten> Vipin Kumar: 1. breadth of interests, not just data mining and machine learning 2. so much in common among people working with very different scientific disciplines 3. we need more success stories to change perception that computer science is a service

<DavidJensen> Journals are a carefully worked-out balance between cooperation and individual reward. Social computing for DI is a whole new realm where we'll need to work out the incentive structures.

<PhilBourne> unsatisified with helping biologist of a very narrow problem - happier working across disciplines

<PhilBourne> challenge is to find teh right people to work with while thinking of teh problem at the same time

<PhilBourne> the social phenomenon is changing our lives

<PhilBourne> Need still better methods to connect people - how can we utilize teh social perspective

<Karsten> Yan Liu: 1. not satisfied solving focused problem with narrow focus -- trying to help the broader community 2. difficult to find scientist willing to work with 3. take advantage of the social revolution to connect the right people, define problems

<DavidJensen> Very interesting point from Yan Liu: Social computing could help crowdsource data gathering for sociology of science.

<PhilBourne> Carla: impressed with preparation of workshop

<LizBradley_> Carla: learned about new systems

<PhilBourne> learnt alot about new systems

QED: quasi experimental design

<PhilBourne> qed

socia computing interesting

<PhilBourne> emphasize social computing as learning a lot

<Yolanda> Carla 1) lots of new information even before workshop

<PhilBourne> a few people mentioned inference and reasoning - not enough

<Yolanda> 2) importnace of social computing

formal reasoning not mentioned as much as expected

<Loren> I forgot to say: I think there's still a tension between problem-centered and technology-centered research.

<PhilBourne> bringing human computation as a way to do reasoning

brinign human computation as a way of doing reasoning, coupling machines with human

<Yolanda> 3) human computation as a way to do reasoning (social computing)

Phil

<Karsten> Carla Gomes: 1. Great workshop organization, gather input ahead of time 2. learned a lot about other systems, new definitions -- especially social computing 3. surprised not too many people mentioned reasoning and inference 4. human computation as a way of doing reasoning is an interesting concept, coupling with machines using the strenghts of each

(cynic)

<DavidJensen> One of the outcomes of studying the foundations of discovery informatics may be a far better understanding of scientific reasoning in general.

<LizBradley_> Phil 1: absurd that we need discovery tools to discover discovery tools

absurd that need discovery tools to create discovery tools

<LizBradley_> tools are not advancing very quickly

<LizBradley_> integration of data is the key first step

<LizBradley_> Emphasis on understanding failure is good

<Kerstin> + integration of data, processes and people

Raul: I learned that way-more scientists use discovery tools than there are computational scientists who can hand-hold them.

<LizBradley_> Emphasis of success of cyberinfrastr overstated

<LizBradley_> Take home point: we are at a tipping point

Raul: Science needs "TurboTax-style" discovery tools that (1) understand scientists's workflows; (2) are expert, automated guides, and (3) are really, really simple to use.

<Yolanda> Phil: success of CI may have been overstated

<LizBradley_> Catalyst often comes from something that we don't expect

<Yolanda> Phil: we are at a tipping point, after this meeting even more convinced

cognition may be changing more than I realized

<Yolanda> Phil: role of cognition research

simplicity, usability, reward will always rule what scientists decide to adopt

<DavidJensen> I was happy to see abductive inference in some of the introductory slides — it's an under-appreciated and fundamental description of much scientific reasoning.

<LizBradley_> Simplicity usability reward will always rule whether scientists adopt something

Turbotax will not quite do it for scientists.

Pirates of Silicon Valley is a great movie

Pat

<LizBradley_> Pleased that induction from large data sets didn't dominate

Pleasantly surprised that abduction from large data sets was not so dominant

<PhilBourne> pat: not obsessed about big datasets induction

<PhilBourne> People not aware of success stories

don't know about long history of work in discoveryin science... wrote survey paper on 8 success stories back in 2000

missing: search metaphor

<PhilBourne> Teh search metaphor was missing - what is teh space and how do you constrain it

<LizBradley_> Study discovery in the context of the whole scientific enterprise

You don't want to study discovery by itself but in context of entire scientific enterprise

Chris

<PhilBourne> Study discovery in the context of teh full scientific enterprise

nice to see so many people interested in this

fragmentation in comp sci is worse than thought

<PhilBourne> Chris: fragmentation in teh computational sciences worse than I thought

<PhilBourne> Chris: NSF truly thinking about the problem

teams should be interdisciplinary

53

<PhilBourne> 5-7 years from a good solution

<DavidJensen> Hod's point about "garage science" raises an interesting thought for me: Can social computing and DI create a *better* incentive structure for research, so that we can massively broaden who can meaningfully contribute to important science.

domain scientists should be on team to create useful tools

Steve

<PhilBourne> Work with a scientist not the science in the broader sense

Feels like kid invited to Santa's workshop

<PhilBourne> Steve: Santa's workshop

<ChrisS> 1. Fragmentation in the computational sciences even worse than I thought

impressed with thoughtful automatization attempt, thoughtful thinking of how to use CS thinking in science

<PhilBourne> Thoughtful automation and how to use computational ability in science

delighted that didn't turn into BIG science, data

<ChrisS> 2. I thought all of NSF already knew that domain people need to be on the teams that build things meant to be used.

ddin't know the fascination with models

<PhilBourne> Fascination of models was not expected

study of science: people who best benefit don't get to see it as much

<ChrisS> 3. Badges (public acknowledgement in labelled form) and computer-supported competitions for data analysis and reviewing is going to be the wave for the future.

<PhilBourne> What is teh takeaway as I move into my craft

discussion between science and discovery was interesting

TurboScience

could be in a high school in 10 years

<PhilBourne> TurboScience - likes that idea- good be in high school in 10 years

collaborations, human/non-human, how can we collaborate with tools better

broadern participation in science through social computing

more people = more energy in science, should be more visible

<DavidJensen> Having a scientist on your project and trying to solve his or her problem is only *one* model of how to do good work in DI. For example, you could work on a problem that has been previously (and well) defined by another community. There may be other successful approaches. Let's not become a monoculture.

<PhilBourne> Very excited by broadening participation in science - more energy comes from more people - science shoudl be a more visibkle enterprise

alex

metrics for discovery

discovery is a diverse concept

many issues are more sociological than technological

<Kerstin> + 1 social aspect much bigger hurdle than technical in many cases

<PhilBourne> Alex: Metrics for discovery is hard discovery is a diverse subject teh human factor is still too large pleasantly surprised by amount of understanding and belief in citizen science

lot to do on the front of models

simplicity is good, e.g. user interfaces

didn't hear a killer app

<PhilBourne> takehome - simplicity of user interfaces - no killer app

education will drammatically change in next 10 years

Cecelia

<PhilBourne> Education will dramatically change in teh next 10 years and we can not be isolated from that

<PatLangley> Many people are unaware of the long history of work on computational discovery in scientific domains, and of the many success stories, so we need to do a better job of advertising.

people approaching people from different angles

importance of human, user-centric design

<Loren> Just came across my twitter feed: An online social network for "pre-submission" peer review of scientific articles: http://news.sciencemag.org/scienceinsider/2012/01/online-social-network-seeks-to.html

<PhilBourne> Cecilia - pleasantly surprised about teh recognition of teh importance of teh human design

<PatLangley> I was surprised that the metaphor of heuristic search was generally absent from the discussions. \

don't know about models, but have built 2 dozen tools for scientists

<PhilBourne> Learnt about the importance of models

domain scientist should be on a team of builders? usually the other way around

Liz

17, 37, 46

command line is terrible interface

<PhilBourne> Command line is a worse interface design than I ever thought

Huan

so many definitions of discovery

<PhilBourne> Huan: so many definitions of discovery

<LarryHunter> Cecelia says in her world, scientists say, Hire a computer scientist instead a graduate student? Why would anyone ever do that? Biology will eventually become that way, but it isn't yet.

<PhilBourne> and ways to discover

so much interesting social computing, expectation of social computing

<PhilBourne> so much expectation of social computing

<PhilBourne> finding global findings from local knowledge

social computing may be able to help discover global knowledge from local knowledge in a tangible way (learned from neighbor)

More people are interested in this area than he thought

<Yolanda> social computing: a way to find global knowledge

<LizBradley_> What I'll do differently: rather than just go work with scientists, I'll think hard about it first: read some papers, do some planning.

socialogy of science, ML, social computing

<PhilBourne> David: concrete ways to work together was evident

lack of whole scientific areas using primitive tools for discovery informatics

<LizBradley_> Did not know that so many people cared so much about the different definitions of "discovery," "models," and so on.

<PhilBourne> Surprised that there are tools for specific examples but not more generic

equivalence classes that may not line up with scientific fields that could be useful for generalization

<CeciliaAragon> I was pleased by the general understanding of the importance of user-centered design in this community.

Andrey

surprise on focus on success stories, could learn from the failures

surprised at how different we are

<PhilBourne> Andrey: surprised how different we are and how we think about research differently

how one discipline thinks about another one

Hod

<PhilBourne> How group dynamics effects what we come up with - would it be different if done remotely?

<PhilBourne> Whole topic is a meta scientific problem

this is a meta scientific problem, surprised that there haven't been more of these meeting. Lots of basic things still aren't nailed down.

Leverage to be gained if automate, make this process more efficient. Should be high priority

<PhilBourne> Hod: so much leverage if we can automate and make processes more efficient

How non-linear the scientific process is.

<PhilBourne> Hod - teh scientif cprocess is soooo non-linear

<CeciliaAragon> Having built a couple dozen systems for scientific collaborations without knowing much about the importance of models, I was excited to learn there's so much work in this area, and am looking forward to using models in my own research.

<PhilBourne> The scientific process is stocastic

process of discovery is not rational

<PhilBourne> The discovery process is open ended and exploratory

garage science, democratization of discovery: NSF can get in the near future by letting people in garages do research with much smaller budget

<PhilBourne> The democratization of discovery - people on an island arguing that the tsunami is coming

<helenadeus> Hod: we are like people on an island discussing whether a tsunami is coming

<PhilBourne> The research enterprise is going to change

problem 42

<CeciliaAragon> It was interesting to see that people kept saying, "a domain scientist should be on a CS project team." Where I come from, scientists generally need to be persuaded that a computer scientist should be on the team to build the code instead of a grad student.

<PhilBourne> Finding the interesting non-trivial thing in teh data - 42

finding the interesting non-trivial thing in the data

Google search is an extension of how people think

<PhilBourne> Google search is an extension of how people think -

tools are not used because they do not add value by producing something new

<PhilBourne> The tools dont add value - they do not allow us to discover something new

distinction between discovery and informatics is really important, should focus on discovery

shouldn't make a TurboTax for science, companies can do this much better

<LarryHunter> +1 for the a successful tool will be one that when a scientists enters her data into the tool and it comes back with something interesting to the scientist in the top one or two results. It's a waste of time to make TurboTax for scientist.

<PhilBourne> trying to compete with industry is a mistake - companiew will do a better TurboScience tool

<LarryHunter> Actually +10 for Hud's point about returning something interesting from the data

<CeciliaAragon> I really disagree that a company can make tools for scientists better - there's not a big enough market for them to want to do this.

Kerstin

discovery of new knowledge in what we have, producing somethign as easy to use as Google but more powerful

<PhilBourne> Kerstin: As useful as Google but across data and other scientific matter

surprised to see so much focused on the long-tail of science

<PhilBourne> So much focus on long tail surprising - used to teh other way around focus on big data

maybe due to composition of workshop

<CeciliaAragon> It's a mistake to assume that the private sector does everything better and more efficiently than anyone else - that's only true if there's a reasonable chance of profit. For non-profit areas, or ones that have only long-term returns, the public sector does better.

problems getting access to real scientists

<PhilBourne> Problems in getting access to real scientists a problem

Cecelia has to fend them off, is there a way of her sharing users and us sharing technologies?

<PhilBourne> How can views and technologies be better shared?

Miriah

<helenadeus> +1 to cecilia's comment

<PhilBourne> A marketplace for informatics

visualization is young but growing, nice to hear its importance in discovery informatics

<PhilBourne> Pleased taht viz research bought to teh front

<helenadeus> it's out job to figure out the prototypes that are useful,only then will industry take it on (and we should let them)

usability, user-centered design

<LarryHunter> -1 to Kerstein for saying that biologists are focused on small science. Biology has been increasingly focused on larger and more expensive and "grander" challenges. The Democratization ("garage") biology is a complement to big biology (not a replacement) as extremely powerful tools become cheap. Note that decades after the introduction of the cheap computer, expensive computer science research has not gone away.

<PhilBourne> knowledge representation resonates

lots of people, lots of different aspects: knowledge representation

Haym

53: galaxy zoom, fold-it

<PhilBourne> Haym - GalaxyZoo and Foldit - we got it straight away

<PhilBourne> excited about teh thinking of data and models

interplay between data and models, process: routine thing in DM, pleased to see this here

<PhilBourne> Pleased about the long tail

long-tail is important... complex data is also interesting

dark science

<PhilBourne> Pleased about the focus on people

people: user-centric design, citixzen science, human bottleneck

will

<PhilBourne> Will: oh

<CeciliaAragon> The problem with saying "human bottleneck" is that it goes back to the old model of thinking that humans are the limiting factor or at fault... I prefer the term "impedance mismatch" between human cognition and computation.

multiple iterations of cartoonmodel of science

prefer the quantum version

<PhilBourne> Will: multiple iteration of the scientific process - prefers quantum version

<PhilBourne> Will: many aspects of science do not work with this model

lot of people are interested in incentivizing use of science/tools

have to build in from the front

<PhilBourne> Will: incentivize use of science and tools - have to build in from teh front - user must be in at the beginning

<LarryHunter> +1 to Cecilia's comment about disagreeing with the "human bottleneck" idea. *All* interesting science comes out of people, and probably will for some time.

<PhilBourne> Think about what the incentives might be

<PhilBourne> What is important to teh scientist must come first

didn't see a lot of talk about social/political forces in science

<PhilBourne> Not much talk about the social and political aspects of science

break down political and social barriers?

<PhilBourne> breakdown teh political barriers not discussed much

<PhilBourne> eg problems of peer review - use technology to change teh structure

<Karsten> yolanda & haym -- saw the whole irc stream but had trouble posting, will e-mail notes.

<DavidJensen> Charge to the breakout groups is to identify exciting research that would be enabled by DI. However, we should focus on things that DI would uniquely enable. What will *uniquely* happen with DI?

<helenadeus> David - categories to be discussed in the breakouts: efficiency, cover, participation, education and basic knowledge about the scientific process

<DavidJensen> To expand...

<DavidJensen> Efficiency — More efficient discovery of what is currently discovered

<DavidJensen> Coverage — Extend what is discoverable beyond what can be discovered currently.

<DavidJensen> Participation — Increase the number of people who can participate in science because of broad availability of tools, data, and collaboration opportunities

<DavidJensen> Education — Increase the availability of actual scientific processes and information to students

<DavidJensen> Knowledge — Increase understanding about the process of scientific discovery because of data collection and basic research on the process itself

<helenadeus> if we want to make an investment in science, we have to understand how we can change it

<Karsten> .com

This is scribe.perl Revision: 1.136  of Date: 2011/05/12 12:01:43  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/Daviv/David/
No ScribeNick specified.  Guessing ScribeNick: susan
Inferring Scribes: susan

WARNING: No "Topic:" lines found.


WARNING: No "Present: ... " found!
Possibly Present: Alex AlexanderSchliep Andrey Carla CeciliaAragon Chris ChrisS ChrisS_ David DavidJensen Haym Hod Huan HuanLiu_ Karsten Kerstin LarryHunter Learning LizBradley_ Loren PaoloCiccarese PatLangley Pat_ Phil PhilBourne QED Raul Steve Will Yolanda andrey_rzhetsky_ helenadeus jls missing pat people
You can indicate people for the Present list like this:
        <dbooth> Present: dbooth jonathan mary
        <dbooth> Present+ amy


WARNING: No meeting title found!
You should specify the meeting title like this:
<dbooth> Meeting: Weekly Baking Club Meeting


WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Got date from IRC log name: 03 Feb 2012
Guessing minutes URL: http://www.w3.org/2012/02/03-diw2012-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report