Vacuum - Edward Vielmetti in Ann Arbor, Michigan 48104

Kanban; or, management by post-it note

I wrote a few years back about a mini-kanban that John Hritz had put together to remember to refill washer fluid. There are a lot of colorful sticky notes around my workplace these days, so I thought I'd read up on how kanban-inspired systems are being used for process management in the software development and other non-manufacturing worlds.

Some inspiring examples (with post-its to illustrate):

Corey Ladis on Kanban systems for software engineering:

A perfect state of flow may be very difficult, or at least uneconomical, to achieve in a robust product development process. But we can get pretty close with a well-tuned kanban pull system. We have managed to combine most of the flexibility of craft production with most of the control of a pipeline. Work-in-process is limited, and cycle time can be managed. Most importantly, it is a highly transparent and repeatable process with all of the right conditions for continuous improvement.

oh, wait a second, this is starting to be spooky: Jim Benson's notes on the new people at his Modus Cooperandi:

Corey Ladas - Corey has been a proponent of iterative and evolutionary design methods since the early 1990’s, and was an early practitioner and vocal promoter of Agile methods at Microsoft. Corey began collaborating with David Anderson at Microsoft in 2004, united by a common interest in the application of Lean, Theory of Constraints, and Statistical Process Control methods to software development. In 2007, Corey joined David at Corbis to implement kanban systems for the development of enterprise IT projects.

Time to talk to Jim...

Technorati Tags: kanban, management

Google quality score rating guidelines leaked

Google Blogoscoped has surfaced a leaked Google Quality Rater Guidelines which purports to show the ranking scale that internal googlers use to assess the quality of search engine results pages. It appears that the document is gone now from the site that hosted it (hm, time to dig through the firefox cache...)

Aside from some insight into what sounds like a day-in, day-out task of reading search results and noting what's wrong (q.v. sisyphus), it gives a sense of the human level of input needed to put the "right" answer top of page.

Abstracted from the text:

Queries are said to be one of three types: navigational (search for known item), informational (research for a topic), or transactional (desire to purchase). There's a set of expectations for each search type; if you type "IBM" the goog wants you to get to IBM.COM on the top of page one. Pages are quality scored as vital, useful, relevant, not relevant, off-topic, didn’t load, foreign language, and unratable,

Technorati Tags: google, search, quality, goog

AATA proposed route changes for Fall 2008, preliminary analysis

The AATA has announced proposed changes to the schedule for this fall. New maps and schedules are not out yet, but there are descriptions of the changes. Here's a breakdown of what wins and what loses with the change.

There's more information and discussion of this on:
- Arbor Update
- Get Downtown
- Ann Arbor News
- Ann Arbor Is Overrated
links to follow when they all get around to writing about it

The short summary, I think, is that the bus system is shifting its service to improve route frequency to the west side of town, cutting service on routes that have low ridership, and shaving a trip here and there. I don't know the total number of service hours before and after (no data yet).

Notably missing from this set of changes is any kind of express bus Ann Arbor to Ypsilanti routing (the mythical 4X route), and any routing that goes directly down Maple/Stadium (the mythical "east-west connector" which could go Sky High to Arborland in one straight shot).

Who wins:
- Route 8 is every 15 minutes! The biggest win
- Route 12UM (Miller) adds service to Sky High, 15 min service on Miller Rd
- minor tweaks to the 4 to speed the journey
- one extra 4 afternoon bus as far as Arborland
- late night (10:45) Blake to Pioneer on the 7

Who loses:
- late trips on the 1U
- evening service on the 3
- latest trip Ypsi-Ann Arbor on the 5
- 12UL goes way (replaced by improved 8 service)
- mid-day 13
- mid-day 14
- summer evening, sunday 15
- 16 changes which I can't parse which probably affect shopping trips

More as I get maps.

Novelty and collective attention / Huberman and Wu

If you wait long enough, it won't be novel, and thus it won't be worthwhile to pay attention to. Thus by incorporating a delay loop into your news consumption you'll miss a bunch of transient spikes of things that are no longer newsworthy.

http://www.pnas.org/cgi/content/abstract/104/45/17599

Novelty and collective attention

Fang Wu and Bernardo A. Huberman*

Information Dynamics Laboratory, HewlettPackard Laboratories, Palo
Alto, CA 94304

Edited by Harry L. Swinney, University of Texas, Austin, TX, and
approved September 14, 2007 (received for review May 25, 2007)

The subject of collective attention is central to an information age
where millions of people are inundated with daily messages. It is thus
of interest to understand how attention to novel items propagates and
eventually fades among large populations. We have analyzed the
dynamics of collective attention among 1 million users of an
interactive web site, digg.com, devoted to thousands of novel news
stories. The observations can be described by a dynamical model
characterized by a single novelty factor. Our measurements indicate
that novelty within groups decays with a stretched-exponential law,
suggesting the existence of a natural time scale over which attention
fades.

(via dragomir radev)

Technorati Tags: attention, novelty, digg

social media metrics / quality without a number

“To seek the timeless way we first must know the quality without a number. There is a central quality which is the rooted criterion of life and the spirit of man, a town, a building, or a wilderness. This quality is objective and precise, but it cannot be numbered.”

Brian Kerr channeling Chris Alexander for a new century and an old industry.

Technorati Tags: qwan

quantitative measures of usability (or, measuring measurability)

I did the Google search for "quantitative measures of usability", to see if that line of thinking would get me something useful. Here's some explorations around those search results, and some reflections on that line of reasoning.

CHI: single usability metric

There is a site "measuring usability" which unsurprisingly ranked high on the list. It introduces the novel concept of a "single measure of usability", which looks like the sum of a bunch of individual metrics all mushed together into a single number. (Sauro and Kindlund, CHI 2005). It measures task completion times, task time, satisfaction and error counts, and is a formalized version of what looks like an original Excel spreadsheet.

From an electronic commerce point of view this is a promising number, but an expensive one to compute - from web server logs you can impute task completion and task times, and maybe get some sense for errors, but satisfaction is hard to grind out of a server log. It comes from the CHI tradition of simple models of user behavior assessed from small data sets. But it's a start.

Patents: quantitative measures of qualitative usability

There's a patent filing (Ed Chi, Christopher Olston) with this initial claim:

A system for determining a quantitative measure of qualitative usability of related Web pages, comprising: stored Web pages that each include at least one hyperlink referencing and proximal cues relating to distal content included in another Web page; a stored information goal identifying a target Web page; an activation network, comprising a directed graph comprising nodes corresponding to the Web pages and arcs corresponding to the hyperlinks, wherein a weight is assigned to each arc to represent a probability of traversal of the corresponding hyperlink based on a relatedness of keywords in the information goal to the proximal cues included in the referenced Web page; and a simulator to evaluate a traversal through the activation network to the node corresponding to the target Web page as a quantitative measure of usability.

There's more (of course, it's a patent, there's a lot more). It appears to be a simulator-driven version of the same PageRank algorithms that Google started out with, except that rather than assuming a random walk through page space it takes into account what Google would call quality scores and what they call "information scent". Chi works at PARC, and this looks like an elaboration of their ScentTrails paper (TOCHI 2003). Note that it's not about task completion as above but more about findability within a search space, and as thus it describes a different problem than above.

More work

I'm sure there's more relevant work in this field, but given that the first words I used to look both unearthed CHI related papers, I'm not sure I have the right language to express the thing I'm really after. so I'll stop...

Bradley Horowitz at U of Michigan campus

Brad Horowitz talked at UMich today. Here are unedited notes....(well, lightly edited).

I was deliciousing links along the way too.

thanks for coming! good to hear the talk.

some notes:

john laird introduces; yahoo speaker series.

bradley grew up in flordia, moved to michigan, "flower king" in livonia
michigan, gymnast in high school. as undergrad in ATL down in the lab
doing all kinds of ramesh jain, terry weymouth, brian shunk (sp??)

from here we lost him to mit media lab to entrepreneur with ramesh,
a bunch of companies, most recently landed at yahoo. vp product strategy.

another indication of us building relationships with yahoo. connections.

--

here to convince you to work with or work at; phd dropout, stay in school kids.
thank my host, incredible day, saw my parents in livonia, overwhelmed by the
work today and its resonance with yahoo. working with us despite us and
around us. a lot to make that easier.

the worst student to stand up in front of you all and profess to have knowledge.

almost failued out a couple of times. came in as CS via LSA and took great books
and history of music and art and not interested in requirements. instead up in
ATL building having fun. took 6 years to get degree and graduate. like george
costanza, scholarship for "rebel bad students", on teh verge of dropping out.

really cool job, now VP Advanced Development Divions, "VP ADD". advanced
in the sense of a scout on the horizon, what's next.

career: looking for first real job, a place to park between startups, fell into yaho0
(after virage...) attitude lasted for a day. quality and caliber of people there. not
a company crowing about how smart they are, "very humble guys". taken by
level of their game. one report, doing multimedia search, image search; launched
video, audio, desktop search. then stumbled across flickr.

helped bring flickr to acquire them. poster child for web 2.0 movement. user-generated
content, rich interfaces with ajax.

runs speaker series for yahoo, brought in Negativland band for IP theft

-------

why do you find the internet depressing? you see what people are
actually doing with this stuff, stupid pet tricks or worse. brings out
the worst in people in a lot of ways.

the antidote is the most interesting photos on flickr.

heuristics that separate the wheat through the chaff, through
the lens of personal

"attention is our most precious asset"

flickr - "lowering the barrier to participation"

flickr compared to computer vision; "computer vision is hard";
man plus machine, not man versus machine.

what makes flickr special?

4. flickr services, api, language bindings, encourage 3d parties
to make flickr bettr.

"this is emergent".

"community and themselves monitor that"

--
Edward Vielmetti +1 734 330 2465

Technorati Tags: yahoo, flickr, upcoming

Monetary Theory and the Great Capitol Hill Baby Sitting Co-op Crisis

from Journal of Money, Credit, and Banking February, 1977, as rescued from the Internet Archive.
There is a JSTOR version of this with good formatting, and at least one other copy in the wild.
Thanks to John Hritz for the pointer.

Apologies for the � marks which are a sign of uncertain provenance. In deference to copyright I have included only the first section of the article.

Monetary Theory and the Great Capitol Hill Baby Sitting Co-op Crisis

By Joan Sweeney and Richard James Sweeney

Sole responsibility for the views expressed here is the author�s. In particular, this does not represent a statement of Treasury views. The authors wish to thank Sevn W. Arndt, David Klock, Dennis E. Logue, Eric Olsen, Jean Willett, and Thomas D. Willett for helpful comments.

Joan Sweeney is Mrs. Richard James Sweeney. Richard James Sweeney is deputy director, Office of International Monetary Research, United States Treasury.

Two of Washington D.C.�s most splendid institutions�the Board of Governors of the Federal Reserve System and the Capitol Hill Baby Sitting Co-operative�are currently fighting their own separate battles against the scourge of inflation. Neither seems to be winning.

Whatever the lessons of the board�s experience, the lessons from the co-op�s are clear. (1) The co-op has been increasing its money supply ("scrip") per capita, by running budget deficits, and this has generated inflationary forces. (2) However, the main "commodity" this scrip money buys is baby-sitting time, and the price of baby sitting is constitutionally pegged at one unit of scrip for every one-half hour of baby sitting. Hence, this system of price controls means the inflationary pressure does not drive up the scrip-price of baby sitting, inflation is suppressed, and shortages are found. (3) The political process of rectifying the situation holds little hope. Few members see the problem as fundamentally monetary, but instead believe others are not doing their part in removing the shortages.

For the uninitiated, it may help to know that there are several forms of baby-sitting co-ops. One popular form is the bookkeeping system. In the most rudimentary version, members earn one credit for each hour of sitting, and lose one credit for every hour someone tolerates their kids. A co-op at this stage develops rules�for fairness, usefulness, for expediency�and to make the thing go at all. For example, people want to go out on Friday and Saturday more than on other days. Either there are rules�"If you go out on weekends, you must sit on weekends"�or there are rewards�"Time-and-a-half on weekends." And, of course, there must be rules to keep people from moving away when they�re "down" on hours.

The major alternative to the bookkeeping system, if there are many people involved, is a "scrip" system�the scrip is pieces of heavy paper. In the Capitol Hill Baby Sitting Co-op, a splendid organization to which we belonged for two years, a unit of scrip "pays" for one-half hour of sitting time. There are good reasons for preferring scrip to bookkeeping. An arithmetic bookkeeping mistake will show members as a whole "ahead" or "down" in hours, and the problem can be hard to resolve. With scrip, the hours earned automatically cancel against the hours spent when the sitter is "paid."

The co-op has enjoyed vicissitudes that make Nixonomics look good by contrast. A few years ago the co-op had a recession. Few people felt they could go out but many wanted to babysit. Now there is great difficulty rounding up sitters for all those who want to go out. This is a classic sort of inflationary pressure�too much money (scrip) chasing too few goods (sitters).

Technorati Tags: kids, via:jhritz, economics, money

What does # mean in a twitter post? All about octothorpetags.

Twitter needs tags.
Twitter doesn't have tags right now.
It does have names, and if you precede a name by an @, you can follow that name.

Sometimes you want to share a conversation but not create an ID.
For instance, when San Diego was on fire, the tag #sandiegofire was used in Twitters.
That's a hashtag.
No software uses hashtags right now.
But if you are regular about using it in your posts and people catch on, maybe someone will write code.

The use of #hashtags to encode names of channels is shared by IRC.
For instance, Joi Ito is founder and op at #joiito on IRC, according to his LinkedIn.
Channel names live in the same world as hashtags and tags.

If you were to build software to use hashtags, what might it do?

Wiki + hashtags = autolink to a wiki where the name space was tags.
IRC + hashtags = autoconnect to the channel.
Google Search + hashtags = search for the tag.
Google Adwords + hashtags = display relevant ads for that tag.
Flickr + hashtags = display a page with that Flickr tag
LinkedIn + hashtags = search for someone with that tag in their profile.
Facebook + hashtags = search for someone with that tag in their profile.

So, essentially, a hashtag is a search key into a tag space, marked with a #.
Indeed, you can use any search engine that searches Twitter to search for them.
Or, at least, any search engine that doesn't ignore the punctuation.

For more about hash tags, read Chris Messina (factoryjoe).
His Twitter hashtags for emergency coordination and disaster relief describes #sandiegofire.

Chris didn't think much of #arbcamp, but we'll forgive him for that just this once.
And I don't think much of the name hashtag.

I'm going to call them octothorpetags.

Technorati Tags: arbcamp, hashtags, factoryjoe, octothorpetags

Human Computation: Luis von Ahn at STIET, Thursday Oct 18 2007

STIET Seminar (Socio-Technical Infrastructure for Electronic Transactions)
Thursday, October 18
4-5:30 pm, 1202 SI North, 1075 Beal Ave.

Human Computation

Luis von Ahn

Assistant Professor of Computer Science, Carnegie Mellon University

Students are invited to a pre-seminar meeting at 3:00 with Luis in 2295 SI North. A background paper can be found at http://www.cs.cmu.edu/~biglou/research.html More information is available at http://stiet.cms.si.umich.edu/node/141

Seminar Description:

Construction of the Empire State Building: 7 million human-hours. The Panama Canal: 20 million human-hours. Estimated number of human-hours spent playing computer solitaire around the world in one year: billions. A problem with today's computer society? No, an opportunity.

What if this time and energy could be channeled into useful work? What if people could play computer games and accomplish work without even realizing it? What if billions of people collaborated to solve important problems for humanity or generate training data for computers? My work aims at a general paradigm for doing exactly that: utilizing human processing power to solve computational problems in a distributed manner. In particular, I focus on harnessing human time and energy for addressing problems that computers cannot yet solve. Although computers have advanced dramatically in many respects over the last 50 years, they still do not possess the basic conceptual intelligence or perceptual capabilities that most humans take for granted. By leveraging human skills and abilities in a novel way, I want to solve large-scale computational problems and/or collect training data to teach computers many of these human talents. To this end, I treat human brains as processors in a distributed system, each performing a small part of a massive computation. Unlike computer processors, however, humans require an incentive in order to become part of a collective computation. Among other things, I use online games as a means to encourage participation in the process. In this talk, I will describe my work in the area of Human Computation.

Seminar Speaker Bio:

Luis von Ahn is an assistant professor in the Computer Science Department at Carnegie Mellon University. He is the recipient of a MacArthur Fellowship, and was named one of Popular Science Magazine's "Brilliant 10" scientists of 2006. His research interests include encouraging people to do work for free, as well as catching and thwarting cheaters in online environments.

Technorati Tags: luis-von-ahn, stiet, google-talks, human-computation, go-urgently, recaptcha, umich