Hackathon/Contest

[saem]

The reading group, as the name would indicate, has been largely hands-
off. A number of us, in a rather informal poll conducted based on
those in my vicinity at the time, were interested in a hackathon.

The rough sketch so far is as follows:

* Random teams from the pool of people -- probably two or three people
per team.
* Have a list of problems determined before hand.
* Have a data set for those problems, the city of Vancouver, and many
others around the world are offering up large data sets to play with.
* Allow some time for the teams to get to know each other and figure
out what problem they wish to tackle, with what technologies, and let
them setup and environment (we don't want to be futzing with things,
and reading docs).
* We take one evening and hack away.
* Likely at a later date, due to the interest of time, we can review
the entries.

So, first off what's the level of interest?

[chuck clark]

+1

Thanks for running with the outcome of our informal poll last night Saem.  I'd definitely like get a little more hands on and be able to do some hands-on exploration.  

I can offer up the space and WiFi at the Pulse Energy offices.  I can also arrange to have some solid and liquid nourishment provided by Pulse.

Would we agree to a problem and then look for each group to solve it using a different set of technologies and compare those solutions similar to coding katas?  Or would each group pick a technology solution set AND a unique problem?

I like the idea of looking at some city data, Vancouver of course would be interesting.  If anyone is interested in solving a problem using energy data we have a dataset used for demos which could be provided.

thanks,

chuck

[Tavis Rudd]

+1

On Sep 21, 9:22 pm, chuck clark <ccl...@ziclix.com> wrote:
> +1
>
> Thanks for running with the outcome of our informal poll last night Saem.
>  I'd definitely like get a little more hands on and be able to do some
> hands-on exploration.
>
> I can offer up the space and WiFi at the Pulse Energy offices.  I can also
> arrange to have some solid and liquid nourishment provided by Pulse.
>
> Would we agree to a problem and then look for each group to solve it using a
> different set of technologies and compare those solutions similar to coding
> katas?  Or would each group pick a technology solution set AND a unique
> problem?
>
> I like the idea of looking at some city data, Vancouver of course would be
> interesting.  If anyone is interested in solving a problem using energy data
> we have a dataset used for demos which could be provided.
>
> thanks,
> chuck
>

[geoff webb]

+1

I too am keen for a little hands on with some of this new fangled technology.

I like the idea of a predetermined set of challenges around some large
datasets, preferably time series data. I also think time-boxing the
session and limiting its scope and breadth of technologies could be
useful in focusing the group.

Sample energy data could be interesting. Especially if it could be
correlated against other city based data - like sporting event
attendance/coverage or other non-related but potentially interesting
events. Here is a fun article about the British need for tea and it's
affect on power consumption -
http://news.bbc.co.uk/2/hi/uk_news/5059904.stm.

Some interesting data sources to further the discussion:
- http://data.vancouver.ca/
- http://stats.oecd.org/index.aspx
- http://data.un.org/Explorer.aspx
- http://mdgs.un.org/unsd/mdg/Data.aspx
- http://robjhyndman.com/TSDL/
- http://www.ngdc.noaa.gov/ngdc.html

and some obvious candidates:
- http://www.data.gov/
- http://www.data.gov.uk/
- http://data.worldbank.org/

g.

[Howard Yeh]

I was at a code retreat in Seattle. Throughout the day, we had 5~6 one
hour sessions doing the conway game of life. Their format was pairing
for an hour, then discussion. The discussion always yielded new ideas
to try in the next iteration. It was interesting to solve the same
problem multiple times, and each time doing it differently.

I'd love to try something like that for noSQL. Maybe we can

1) decide on a dataset
2) decide on 3~4 data storages, one of which is a SQL

For each storage

3) load & map the dataset into the storage
4) see what sort of reporting the storage makes it easy/hard
(maybe we have a pre-determined set of reporting. Average, max.
stddev (all of which ranged by time))

My 2c

Howard

[Jerry Cattell]

+1 with the caveat that I am out of town half of October.

On Tue, Sep 21, 2010 at 9:04 PM, saem <saem...@gmail.com> wrote:

[Ken Pratt]

+1

I think it would be interesting if everyone worked on the same data
set. That would give us the most insight to compare & contrast the
platforms afterward.

I also like Howard's idea of deciding on 3-4 storage systems,
including SQL if one team is willing to take it.

-Ken

[saem]

I have to say so far Howard's idea is the most compelling. Harvey and
I talked about this at lunch today and he put forth a few ideas I
won't steal his thunder. Along with Geoff and I brainstorming at work.

The idea of rapidly iterating and trying out various ideas in small
teams so we can get a breadth of solutions is very appealingly. I'm
taking this roughly as 'voting' and we can work it out at the next
gathering. From there we can go forward with dates, problems and data
sets.

[Simon Claret]

+1

Personally I would prefer to work on unique problems, although I can
understand why some might want to see solutions to the same problem
side-by-side.

Is anyone interested in forming team erlang?

Simon

[David Dossot]

I'm old and tired hence would be happy to stay on familiar ground and join team Erlang :)

D.

[Dethe Elza]

On 2010-09-23, at 8:14 AM, David Dossot wrote:

> I'm old and tired hence would be happy to stay on familiar ground and join team Erlang :)

+1

I think I may stay at home and teach my kids Javascript. But y'all have fun. Report any findings.

--Dethe

[Howard Yeh]

> I think I may stay at home and teach my kids Javascript. But y'all have fun. Report any findings.

<flamebait>
why not teach them Logo? don't warp their minds with C-syntax.
</flamebait>

[Dethe Elza]

I want to teach them Javascript with Scratch syntax.

I love Logo, but they can't write games for the web using it.

--Dethe

[Ken Pratt]

I suggest that we pick teams based on data stores, and then each team
can decide on an appropriate implementation language (since most
stores have decent libraries for most languages, and this event is
about the stores and not the languages).

If we're randomizing, cool. If not, I nominate a Riak team.

-Ken

--
Ken Pratt
http://kenpratt.net/