55 Matching Annotations
  1. Last 7 days
  2. Jul 2021
  3. Jun 2021
    1. One thing that should be learned from the bitter lesson is the great power of general purpose methods, of methods that continue to scale with increased computation even as the available computation becomes very great. The two methods that seem to scale arbitrarily in this way are search and learning

      This is a big lesson. As a field, we still have not thoroughly learned it, as we are continuing to make the same kind of mistakes. To see this, and to effectively resist it, we have to understand the appeal of these mistakes. We have to learn the bitter lesson that building in how we think we think does not work in the long run. The bitter lesson is based on the historical observations that 1) AI researchers have often tried to build knowledge into their agents, 2) this always helps in the short term, and is personally satisfying to the researcher, but 3) in the long run it plateaus and even inhibits further progress, and 4) breakthrough progress eventually arrives by an opposing approach based on scaling computation by search and learning. The eventual success is tinged with bitterness, and often incompletely digested, because it is success over a favored, human-centric approach.

  4. May 2021
  5. Mar 2021
    1. Occasionally, like with search engines, #2 occurs because the incumbents gain massive economies of scale (classic Microeconomics), where by virtue of their being large, the cost to produce each incremental good or service at scale becomes much lower.
  6. Feb 2021
  7. Jan 2021
  8. Dec 2020
    1. SoulCycle was never built to be for the masses. Keeping people out was, it seems, just as important to the business as loyal riders. The bigger SoulCycle got, the less desirable it became. The less desirable it became, the less people had tolerance for the culture it fostered. The minute the company became mainstream, the magic dissolved. It’s impossible to scale exclusivity.
  9. Oct 2020
  10. Sep 2020
  11. Aug 2020
  12. Jul 2020
  13. Jun 2020
  14. May 2020
  15. Apr 2020
  16. Aug 2019
    1. tate it.Whenever an annotation was added to a Madison document, a few technical features helped to further facilitate conversation. First, the document’s sponsor was automatically notified of a new annotation. Second, the annotation also appeared in-line as marginalia that could be responded to, liked, or flagged by others. And third, the annotation was displayed as a “comment” along with others at the end of the document. This process was described as “the future of crowdsourced legislation,” and illustrated how social and collaborative annotation could contribute to and improve civic life.Among noTroy Hicks1 week agoIt seems that these technical features were ones that, I am assuming, where only known and used by a very few of the users. Again, speaking to power and access, what does that mean for the kinds of democratized annotation experiences that we aspire to? How is this (entirely) dissimilar from conversations on social media, perhaps even off-putting or inaccessible to average users?

      Or additionally consider the vast amounts of un-curated noise that annotations may make in instances like these when they hit larger scale. How can these systems better delineate the authority of the individual authors?

      As a foil, consider how often people may read the several thousands of comments on a particular New York Times article? How many readers delve into these conversations and interact with them—particularly when they aren’t moderated or are overpopulated by trolls? We need better UI to indicate those annotating with some authority (or provide their background and expertise) or who may even be the original author responding to questions.

  17. Mar 2019
    1. Scaling In: The Community Dimensions of Innovation

      Great reading for answers to the question "yes, but does it scale?"

    2. To solve the problem of ‘scaling up’ requires ‘scaling in’ –by this we mean developing the designs and infrastructure needed to support effective use of an innovation.

      On "scaling-in" rather than "scaling-up".

    3. special thanks to Nate Angell

      Whoa! Thanks Robin! I don't remember what I might have contributed, but super-honored to be mentioned in such an awesome work!

  18. Jan 2019
    1. For large-scale software systems, Van Roy believes we need to embrace a self-sufficient style of system design in which systems become self-configuring, healing, adapting, etc.. The system has components as first class entities (specified by closures), that can be manipulated through higher-order programming. Components communicate through message-passing. Named state and transactions support system configuration and maintenance. On top of this, the system itself should be designed as a set of interlocking feedback loops.

      This is aimed at System Design, from a distributed systems perspective.

  19. Oct 2018
    1. 51 x 36 cm

      This drawing is fairly small, but many of Voigt's pieces are quite large - certainly as tall as she is. Scale does seem to have a place in the field of objects For Boredom is concerned with. Not all written works are long, but a lot of them are decidedly overwhelming, and certainly some of the novels are, too. Voigt’s are a bit different because you can visually take it all in at once, but still, they are large-scale and highly detailed.

    1. All  hypotheses  confirmed  •Engaging  in  discussion  leads  to  more  correct  answers.  •  The  bonus  incentive  leads  to  more  correct  changed  answers.  •The  participants  have  substantive  discussio

      Interesting finding based on MTurk experiments. Discussion and incentive matter.

    2. MOOC  Collaboration  Today  •Forums  •Really  Q&A  Tools  •Low  participation  •Participants  do  well:  correlation  or  causation?  •Informally  Organized  Groups  •Google  Hangous,  Facebook  groups,  in-­‐person  meetings  •Formal  Project  Groups  •NovoEd  •Peer  Assessment  (anonymous,  asynchronous)  •Kulkarni,  Klemmer  et  al.  TOCHI  2

      These activities are arguable cooperative. Also, they are mostly defined by the instructor.

  20. Aug 2018
    1. Contemporary thinkers from a wide range of fields arrived at an understanding that recognises the implication of our past in the present; in other words, that our personal and social history forms an ineradicable part of us. I can find no good reason why we should exlude our biological and cosmic past from the acceptance of this general principle.

      Adam contests Gidden's perspective on time-scale because he does not integrate biological or cosmic evolution into the influence that personal and social history can have on how people experience the present through the past.

  21. Jun 2018
    1. 51 x 36 cm

      This drawing is fairly small, but many of Voigt's pieces are quite large - certainly as tall as she is. Scale does seem to have a place in the field of objects For Boredom is concerned with. Not all written works are long, but a lot of them are decidedly overwhelming, and certainly some of the novels are, too. Voigt’s are a bit different because you can visually take it all in at once, but still, they are large-scale and highly detailed.

  22. Sep 2017
  23. Aug 2016
    1. Page 6

      Borgman on the importance of scale in information retrieval. It's an interesting question for the humanities not only does large-scale introduce new methods for example just reading it also makes traditional methods more difficult EG challenges close reading. It is not enough to say (as color and others do) that they don't like distant reading. They also need to say how they propose doing the reading in a million book environment.

      data and information have always been both input and output of research. What is new is the scale of the data and information involved. Information management is notoriously subject to problems of scale [bibliography removed]. Retrieval methods designed for small databases declined rapidly ineffectiveness as collections grow in size. For example a typical searcher is willing to browse a set of matches consisting of one percent of a database of 1000 documents (10 documents), maybe willing to browse a 1% set of 10,000 documents (100), rarely is willing to browse 1% of 100,000 documents (1000), and almost never would browse 1% of 1 million or 10 million documents.

  24. Jul 2016
    1. Page 16

      guessing domain names declined in Effectiveness as the ww.w grew in size, which is another example of the scaling problems of information retrieval.

      I wonder if this is true. Did people really get domain names?

  25. Mar 2016
    1. And this leads me to another thought- it seems like in our field there is this desire to go big, to scale, to teach hundreds of thousands, to affect an entire sector. Scale at the dimension is really only achieved by a process of mass duplication where the level of heart-felt connectivity is probably low.
  26. Dec 2015
    1. It is possible to achieve a more humane and personal education at scale

      Important claim, probably coming from the need for reports which answer the “But does it scale?” question.

  27. Oct 2015
    1. you can never slow down too much. It’s impossible todisconnect. Right now I’ve got a ferry to catch from Swartz Bay to the GulfIslands, and Tony has a plane to catch at the airport. The idea of islandtime is all about trying, this is the keyword,trying, to slow down.

      Again the idea of being "in time" as a scale. People living with the island state of mind must still take outside influences into account, such as flight departures.

  28. Jan 2014
    1. The creation and exploitation of large-scale quantitative atlases will lead to a more precise understanding of development.

      large-scale quantitative atlases lead to more precise understanding

    2. Just as comprehensive datasets of genomic sequence have revolutionalized biological discovery, large-scale quantitative measurements of gene expression and morphology will certainly be of great assistance in enabling computational embryology in the future. Such datasets will form the essential basis for systems level, computational models of molecular pathways and how gene expression concentrations and interactions alter to drive changes in cell shape, movement, connection, and differentiation. In this review, we discuss the strategies and methods used to generate such datasets.