176 Matching Annotations
  1. Jan 2024
    1. What made the deal so unusual - and sparked concern among exhibitors - is that 2929 plans to distribute the projects simultaneously in the theatrical, home video and cable arenas.

      https://www.theguardian.com/film/2005/may/27/news1

      In 2005, 2929 Entertainment signed a distribution deal with Steven Soderbergh which was one of the first to suggest distributing projects simultaneously to theatrical, home video, and cable.

  2. Dec 2023
    1. Adler & Hutchinson's Great Books of the Western World was an encyclopedia-based attempt to focus society on a shared history as their common ground. H. G. Wells in his World Encyclopedia thesis attempts to forge a new "moving" common ground based on newly evolving knowledge based on distilling truth out of science. Shared history is obviously much easier to dispense and spread about compared to constantly keeping a growing population up to date with the forefront of science.

      How could one carefully compose and juxtapose the two to have a stronger combined effect?

      How could one distribute the effects evenly?

      What does the statistical mechanics for knowledge management look like at the level of societies and nations?

      link to https://hypothes.is/a/abTT1KPDEe6nqxPx4fXggw

  3. Jul 2023
  4. Jun 2023
  5. Mar 2023
    1. We find that meeting such access needs for the billions in poverty may lead to crossing ESBs unless resources are reallocated from the rich to the poor28, in line with limitarian and sufficientarian justice37,73.
      • Comment
      • The transformation of the economically wealthy will be critical to the future of civilization
  6. Oct 2022
    1. Under a normal distribution, the area would be about 0.003 using the 68-95-99.7 rule.

      Normal rule

    2. However, its tails are thicker than the normal distribution’s, meaning observations are more likely to fall beyond two standard deviations from the mean than under the normal distribution.

      This is what Dr. Herring meant by "fatter tails."

      Below is why.

  7. Aug 2022
    1. Distribution (or place) refers to an organization, or set of organizations, that is involved in the process of making a product or service available for use or consumption by a consumer or business user
    1. modern trade is still in the very early stages of development. The numerous traditional trade outlets (e.g. small groceries, mom-and-pop shops, dukas or souks) remain the biggest segment of the market
    2. visibility in the supply chain remains one of the biggest challenges. As outlets are small, contributing low volume, hardware and software costs are major stumbling blocks. African companies are increasingly assessing mid-tech solutions and identifying the “appropriate technology” for their operation
    1. Generating randomness.

      EVM execution is deterministic. How to account for randomness?

      Pesudo random generator, probability distribution.

  8. Jun 2022
    1. This podcast is also available on Spotify, Apple Podcasts and Anchor. Please subscribe on your favoured podcast provider and leave a review.

      There are actually seven different services that this podcaster has done a huge amount of work to put their content on, ostensibly for the widest discovery, but not a single one of them has a link to the raw audio file to make it easy for one to bookmark and listen to later. Apparently the podcasting silo services have managed to win out over the open web.

      Do we really need to make podcasting this hard on individual publishers? Why can't the publisher just have one location and tell all the aggregators, here's a link to my feed, copy it if you will and want to help distribute my content? In some sense, this is part of what is happening as all seven services are using the same findable source, they're just making it more difficult to jump through all the hoops, which means the small guys end up paying more to do the extra work and potentially lose everything if that one source disappears, closes down, or gets acquired and goes away.

      These sorts of artificial hurdles and problems are what make it so hard to get up and running.

  9. Apr 2022
  10. Mar 2022
    1. Eran Segal. (2021, August 17). Israel data showing the decay of vaccine efficacy over time. Y-axis is cases per 1000 from July 7 to Aug 10, for unvaccinated, and for people vaccinated at different times Cases are higher in those vaxed earlier Despite world-data caveats, this seems quite compelling https://t.co/5aNz48AC8F [Tweet]. @segal_eran. https://twitter.com/segal_eran/status/1427696623988117505

    2. Natalie E. Dean, PhD. (2021, August 17). Real-world data from Israel show a growing gap between the earliest vaccinated (blue arrow) and the recently vaccinated (green arrow) within age groups. Confounding is always a concern (are these groups fundamentally different?) but the magnitude of the difference is notable. Https://t.co/s8pevRbax8 [Tweet]. @nataliexdean. https://twitter.com/nataliexdean/status/1427703094062706691

  11. Feb 2022
  12. Jan 2022
    1. 1.1 Bernoulli distribution

      $$ Y \sim f_{B}(y ; \theta)= \begin{cases}\theta^{y}(1-\theta)^{1-y} & \forall y \in\{0,1\} \\ 0 & \text { otherwise }\end{cases} $$

      $$E[Y]=\theta$$

      $$var(Y)=\theta(1-\theta)$$

    Tags

    Annotators

  13. Dec 2021
  14. Nov 2021
  15. Oct 2021
  16. Sep 2021
  17. Aug 2021
    1. Madhu Pai, MD, PhD. (2021, August 4). 3.5 billion people in🌍 have not had a single Covid 💉 Meanwhile, rich nations are: - Throwing away expired vaccines—Giving booster shots—Offering lotteries to people who are hesitant—Hoarding doses for next year 𝙃𝙊𝙒 𝙒𝙄𝙇𝙇 𝙏𝙃𝙄𝙎 𝙋𝘼𝙉𝘿𝙀𝙈𝙄𝘾 𝙀𝙑𝙀𝙍 𝙀𝙉𝘿? [Tweet]. @paimadhu. https://twitter.com/paimadhu/status/1422880112387710982

  18. Jul 2021
  19. Jun 2021
    1. There's no official Chrome or Chromium package for Linux don't install it this way because it's either outdated or unofficial, both are bad. Download it from official source.
  20. May 2021
    1. Substack insists that advances are determined by “business decisions, not editorial ones”. Yet it offers writers mentoring and legal advice, and will soon provide editing services.

      Some evidence of Substack acting along the lines of agent, production company, and studio. Then taking a slice of the overall pie.

      By having the breadth of the space they're able to see who to invest in over time, much the same way that Amazon can put smaller companies out of business by knocking off big sales items.

    2. Record labels are another endangered middleman. They have historically taken care of turning a song into a hit, in return for an ongoing share of revenues. But more and more artists are going it alone. More than 60,000 new songs are uploaded to Spotify every day, most by bedroom-based rockstars who can use new online services to handle the logistics themselves. UnitedMasters, a music-distribution platform which bills itself as “a record label in your pocket”, recently raised $50m in a venture-capital round led by Apple. Tools like Splice make recording easier. Companies like Fanjoy take care of merchandise.And financing is getting simpler. One startup, HIFI, helps artists manage their royalties, paying them regularly and fronting them small sums to make up shortfalls. Another, Karat, extends credit to creators based on their follower count. Helped by such services independent artists took home 5.1% of global recorded music revenues last year, up from 1.7% in 2015, calculates MIDiA Research, a consultancy. In the same period the share of the three largest record labels fell from 71.1% to 65.5%.

      The same sort of dis-aggregation and disintermediation that has hit the publishing business is also taking place to newspapers, magazines, and music.

      The question is how to best put the pieces of the pie together in the best way possible. There's probably room for talented producers to put these together to better leverage the artists' work.

  21. Apr 2021
    1. In the coming months and years, we’ll be working to further enable choice for creators, including giving them the power to choose not only how someone wants to create or monetize audio, but also where specific content is able to be consumed, ensuring creators have an opportunity to decide if they are aligned with the platforms distributing their content.

      So this means you're going to use simple, open standards and tooling so that not only Anchor and Spotify will benefit? Or are you going to build closed systems that require the use of proprietary software and thus force subscriptions? Are you going to Balkanize the audio space to force consumers into your product and only your product? Or will producers be able to have a broad selection of platforms to which they could distribute their content?

    1. Kai Kupferschmidt. ‘According to @PEI_Germany about 2,7 Million People Have Now Been Vaccinated with AstraZenaca Vaccine in Germany. Amongst These: 31 Cerebral Venous Thromboses (29 Women) 19 of These Also with Thrombocytopenia Reported 9 Deaths Clearly Germany Has to Change Recommendations for Now’. Tweet. @kakape (blog), 30 March 2021. https://twitter.com/kakape/status/1376859903030071301.

    2. Kai Kupferschmidt. ‘According to @PEI_Germany about 2,7 Million People Have Now Been Vaccinated with AstraZenaca Vaccine in Germany. Amongst These: 31 Cerebral Venous Thromboses (29 Women) 19 of These Also with Thrombocytopenia Reported 9 Deaths Clearly Germany Has to Change Recommendations for Now’. Tweet. @kakape (blog), 30 March 2021. https://twitter.com/kakape/status/1376859903030071301.

  22. Mar 2021
    1. Substack’s nice interface and large community made it easy for content to go viral. And that’s what I wanted. I didn’t need to be paid, but I wanted to get some of my weirder ideas in front of a broad audience. What I’m saying is that Substack suckered me in with the promise of growing my readership, and the bait was that they had so many great writers with huge followings. But now I’m left wondering how many of those huge followings were made possible by payouts from Substack. 

      YouTube's model is certainly more mature, but is very similar. Some very high profile creators get paid very well and act as scions for hoi poloi who also think they can replicate the same system and become rich themselves. The incredibly vast majority will never come close.

    1. ReconfigBehSci. ‘RT @ashishkjha: Over Past Week We Got 11.4 Million Doses into Arms 5.6 Million Were 1st Doses 5.8 Million Were 2nd Doses That’s a Proble…’. Tweet. @SciBeh (blog), 1 March 2021. https://twitter.com/SciBeh/status/1366421544495382533.

  23. Feb 2021
    1. As of today, you can Wishlist OpenTTD on SteamE. Historically, OpenTTD always had a single home from where we distributed the game. We used to be hosted on SourceForge (you know you are old, if you remember that being a thing :D), and slowly moved towards our own self-created distribution methods. These days, we mostly distribute our game via our website. But times are changing, and so is our hair. Over the last few months, we have silently been working to become a bit more visible in the world. Don’t worry, not for reasons you might think: OpenTTD has as many active users as it had in 2007. But more because we no longer think it is the right approach to only distribute via our own website. This became painfully apparent when we noticed other people post OpenTTD on some stores. They are not always updated with new releases, sometimes even slacking behind a few years. And maybe more important to us: we can not guarantee that the uploaded version is unmodified and is the version as we intended. So, instead of fighting it, why not turn around and join them! Why not release our own, verified, builds on those stores! And this is exactly what we have been working on lately. And when I say “we”, a bit ironic to me, I mean the two developers that are around longest (myself and orudge) ;) A while back orudge added OpenTTD to the Microsoft Store. And today, I am happy to announce we will be on SteamE too! Well, we are on Steam, but we haven’t released anything there yet (sorry that I got your hopes up, just to squash them right after :( ). This is partially because of how Steam works, but also because we know we can bring a better experience for Steam with our upcoming release. That brings me to the most exciting news: if everything goes as planned, we will release OpenTTD 1.11 on Steam on the first of April, 2021! And that is not even an April fools’ joke! You can already Wishlist OpenTTD today .. and till we release on Steam, you can find our game via our website ;)
    1. As of today, you can Wishlist OpenTTD on SteamE. Historically, OpenTTD always had a single home from where we distributed the game. We used to be hosted on SourceForge (you know you are old, if you remember that being a thing :D), and slowly moved towards our own self-created distribution methods. These days, we mostly distribute our game via our website. But times are changing, and so is our hair. Over the last few months, we have silently been working to become a bit more visible in the world. Don’t worry, not for reasons you might think: OpenTTD has as many active users as it had in 2007. But more because we no longer think it is the right approach to only distribute via our own website.
  24. Jan 2021
  25. Dec 2020
  26. Nov 2020
  27. Oct 2020
    1. Publishers who aren’t media partners with Facebook, Snapchat and Twitter, aren’t highlighted prominently on these platforms, don’t receive a heads up about new products and never have a direct line to support at these companies.

      Looking at the relationship of authors, book publishers, and even the big 5 publishing companies provides a reasonable model for what all of this looks like down the road. All the publishers are generally screwed if they're reliant on one distributor which they don't control.

    1. “Every Cambodian... including the King has the right to express freely their view.”

      While I like the sentiment here, a lot of the power of the message comes from not only the medium, but the distribution which it receives. Many daily examples of "typical" annotation done by common people are done in a way that incredibly few will ultimately see the message. The fact that the annotations of the emperor were republished and distributed was what, in great part, gave them so much weight and value. Similarly here with the example of the King's blog or Alexandra Bell's work which was displayed in public. I hope there is more discussion about the idea of distribution in what follows.

  28. Sep 2020
  29. Aug 2020
    1. Put another way, a lot of the “low hanging fruit” in the US software market is now gone. Software in the US generally works. And new opportunities get swept up with would-be competitors immediately. If the 90s was about thinking through your build, the 2020s is about thinking through marketing & distribution.

      The low hanging fruit in software markets is now gone in the US. New opportunities get swept up immediately. The 90s were about figuring out how to build it, the 2020s are about figuring out marketing & distribution.

    1. Remote work distributes wealth into the whole system

      I think there is potential to do that, but then you hear cases like how some companies have reduced the pay for remote workers based on the cost of living of the cities that they live in.

  30. Jul 2020
  31. Jun 2020
  32. May 2020
  33. Apr 2020
  34. Dec 2019
    1. So if you create one backup per night, for example with a cronjob, then this retention policy gives you 512 days of retention. This is useful but this can require to much disk space, that is why we have included a non-linear distribution policy. In short, we keep only the oldest backup in the range 257-512, and also in the range 129-256, and so on. This exponential distribution in time of the backups retains more backups in the short term and less in the long term; it keeps only 10 or 11 backups but spans a retention of 257-512 days.
    1. the exponential distribution is the probability distribution of the time between events in a Poisson point process, i.e., a process in which events occur continuously and independently at a constant average rate.
  35. Oct 2019
    1. Install On Air is best application distribution service over the air. We offer Beta App, iOS App and Android App distribution. Don’t waste time and have an easy process for testing builds.

  36. Aug 2019
    1. of power, let’s consider another type of annotation, written by a single author, and in response to a very different though equally important social and political circumstance.PoetryMaha Bali2 weeks agoAt this point, a thought crossed my mind related to Audrey Watters feeling like she did not want annotation to happen on her own website (but is happy to have it done outside her website if people wanted). I wonder if there is any value in unpacking that one here? Perhaps. perhaps not.

      Ultimately Audrey Watters rescinded the Creative Commons license on her website, though I don’t think she ever mentioned specifically why she made that change (nor does she need to publicly state a reason) though it may have had something to do with annotations and/or harassment she experienced at the time.

      I do remember thinking at the time she was looking at those decisions that in some sense by allowing annotations on her site, she was providing a platform and distribution for others to potentially harass her.

      Some pieces of that extended conversation:

      https://boffosocko.com/2017/05/10/un-annotated-by-audrey-watters/

      https://blog.jonudell.net/2017/06/27/annotating-thoughts-on-annotation/

  37. Jul 2019
    1. The initial promise of Web 2.0—that gatekeepers have their power reduced and that “ordinary” users can make media—is still true, even for for-profit firms such as Facebook and Twitter.

      While this is generally true that everyone can now create, the real inequity is the fact that distribution is not equal for all players. We might also ask the question: Should distribution be equal for all?

      Robin Sloan has an "essay" on this topic that mirrors my own long held distribution questions/problems: https://platforms.fyi/

    2. In the years between 2004 and 2012, many media critics proclaimed a promising new mediascape of democratic production and thus democratic organization (Benkler, 2006; Bruns, 2008; Shirky, 2009)—precisely what alternative media theorists had been calling for in previous decades.

      I note here that they mention production and organization, but there is a missing piece of "distribution". In large part, part of the problem with current corporate social media is one of how their content is distributed and the advertising model that drives what sorts of content are distributed.

  38. Apr 2019
    1. We make our branded beverage products available to consumers throughout the world through our network of independent bottling partners, distributors, wholesalers and retailers as well as Company-owned or -controlled bottling and distribution operations — the world's largest beverage distribution system.
  39. Feb 2019
    1. What happened is that Spotify dragged the record labels into a completely new business model that relied on Internet assumptions, instead of fighting them: if duplicating and distributing digital media is free (on a marginal basis), don’t try to make it scarce, but instead make it abundant and charge for the convenience of accessing just about all of it.
  40. Aug 2017
    1. Thus, predicting species responses to novel climates is problematic, because we often lack sufficient observational data to fully determine in which climates a species can or cannot grow (Figure 3). Fortunately, the no-analog problem only affects niche modeling when (1) the envelope of observed climates truncates a fundamental niche and (2) the direction of environmental change causes currently unobserved portions of a species' fundamental niche to open up (Figure 5). Species-level uncertainties accumulate at the community level owing to ecological interactions, so the composition and structure of communities in novel climate regimes will be difficult to predict. Increases in atmospheric CO2 should increase the temperature optimum for photosynthesis and reduce sensitivity to moisture stress (Sage and Coleman 2001), weakening the foundation for applying present empirical plant–climate relationships to predict species' responses to future climates. At worst, we may only be able to predict that many novel communities will emerge and surprises will occur. Mechanistic ecological models, such as dynamic global vegetation models (Cramer et al. 2001), are in principle better suited for predicting responses to novel climates. However, in practice, most such models include only a limited number of plant functional types (and so are not designed for modeling species-level responses), or they are partially parameterized using modern ecological observations (and thus may have limited predictive power in no-analog settings).

      Very nice summary of some of the challenges to using models of contemporary species distributions for forecasting changes in distribution.

    2. In eastern North America, the high pollen abundances of temperate tree taxa (Fraxinus, Ostrya/Carpinus, Ulmus) in these highly seasonal climates may be explained by their position at the edge of the current North American climate envelope (Williams et al. 2006; Figure 3). This pattern suggests that the fundamental niches for these taxa extend beyond the set of climates observed at present (Figure 3), so that these taxa may be able to sustain more seasonal regimes than exist anywhere today (eg Figure 1), as long as winter temperatures do not fall below the −40°C mean daily freezing limit for temperate trees (Sakai and Weiser 1973).

      Recognizing where species are relative to the observed climate range will be important for understanding their potential response to changes in climate. This information should be included when using distribution models to predict changes in species distributions. Ideally this information could be used in making point estimates, but at a minimum understanding its impact on uncertainty would be a step forward.

  41. Jun 2017
  42. Mar 2017
    1. One implication of the naturalness with which we divide cognitive labor,” they write, is that there’s “no sharp boundary between one person’s ideas and knowledge” and “those of other members” of the group.
  43. Jan 2017
    1. To simulate equilibrium sagebrush cover under projected future climate, we applied average projected changes in precipitation and temperature to the observed climate time series. For each GCM and RCP scenario combination, we calculated average precipitation and temperature over the 1950–2000 time period and the 2050–2098 time period. We then calculated the absolute change in temperature between the two time periods (ΔT) and the proportional change in precipitation between the two time periods (ΔP) for each GCM and RCP scenario combination. Lastly, we applied ΔT and ΔP to the observed 28-year climate time series to generate a future climate time series for each GCM and RCP scenario combination. These generated climate time series were used to simulate equilibrium sagebrush cover.

      This is an interesting approach to forecasting future climate values with variation.

      1. Use GCMs to predict long-term change in climate condition
      2. Add this change to the observed time-series
      3. Simulate off of this adjusted time-series

      Given short-term variability may be important, that it is not the focus of the long-term GCM models, and that the goal here is modeling equilibrum (not transitional) dynamics, this seems like a nice compromise approach to capture both long-term and short-term variation in climate.

    2. Our process model (in Eq. (2)) includes a log transformation of the observations (log(yt − 1)). Thus, our model does not accommodate zeros. Fortunately, we had very few instances where pixels had 0% cover at time t − 1 (n = 47, which is 0.01% of the data set). Thus, we excluded those pixels from the model fitting process. However, when simulating the process, we needed to include possible transitions from zero to nonzero percent cover. We fit an intercept-only logistic model to estimate the probability of a pixel going from zero to nonzero cover: yi∼Bernoulli(μi)(8)logit(μi)=b0(9)where y is a vector of 0s and 1s corresponding to whether a pixel was colonized (>0% cover) or not (remains at 0% cover) and μi is the expected probability of colonization as a function of the mean probability of colonization (b0). We fit this simple model using the “glm” command in R (R Core Team 2014). For data sets in which zeros are more common and the colonization process more important, the same spatial statistical approach we used for our cover change model could be applied and covariates such as cover of neighboring cells could be included.

      This seems like a perfectly reasonable approach in this context. As models like this are scaled up to larger spatial extents the proportion of locations with zero abundance will increase and so generalizing the use of this approach will require a different approach to handling zeros.

    3. Our approach models interannual changes in plant cover as a function of seasonal climate variables. We used daily historic weather data for the center of our study site from the NASA Daymet data set (available online: http://daymet.ornl.gov/). The Daymet weather data are interpolated between coarse observation units and capture some spatial variation. We relied on weather data for the centroid of our study area.

      This seems to imply that only a single environmental time-series was used across all of the spatial locations. This is reasonable given the spatial extent of the data, but it will be necessary to allow location specific environmental time-series to allow this to be generalized to large spatial extents.

    4. Because SDMs typically rely on occurrence data, their projections of habitat suitability or probability of occurrence provide little information on the future states of populations in the core of their range—areas where a species exists now and is expected to persist in the future (Ehrlén and Morris 2015).

      The fact that most species distribution models treat locations within a species range as being of equivalent quality for the species regardless of whether there are 2 or 2000 individuals of that species is a core weakness of the occupancy based approach to modeling these problems. Approaches, like those in this paper, that attempt to address this weakness are really valuable.

  44. Dec 2016
    1. An open source infrastructure for a centralized network now provides almost the same level of control as federated protocols, without giving up the ability to adapt. If a centralized provider with an open source infrastructure ever makes horrible changes, those that disagree have the software they need to run their own alternative instead. It may not be as beautiful as federation, but at this point it seems that it will have to do. Tweet

      I'm not sure if this comparison is really working: What if I really take the Signal software, because I have reasons to do so. How can I stay upstream compatible, if "upstream" means the master branch of Open Whisper Systems? There soon will be two communities at least for me: the one that I left behind and the one that went with me to the new installation. But how can they communicate with each other? With the installation of the second instance the "Signal" communication has become distributed in a way except for the two instances cannot talk to each other. As moxie0 says above: "When someone recently asked me about federating an unrelated communication platform into the Signal network, I told them that I thought we'd be unlikely to ever federate with clients and servers we don't control."

    2. This reduced user friction has begun to extend the implicit threat that used to come with federated services into centralized services as well. Where as before you could switch hosts, or even decide to run your own server, now users are simply switching entire networks. In many cases that cost is now much lower than the federated switching cost of changing your email address to use a different email provider.

      There it is again: convenience as the main driver for the ecosystem to develop.

      The "cost" mentioned here is the freedom of not having to send my personal social graph to a server that might belong to someone else tomorrow.

      The two things compared do not fit: Switching networks on the basis of a phone number can be compared to switching similar services with the equivalent of an email address. And changing your email provider can be compared to changing your phone company without being able to take your phone number with you.

    3. If anything, protecting meta-data is going to require innovation in new protocols and software. Those changes are only likely to be possible in centralized environments with more control, rather than less. Just as making the changes to consistently deploy end to end encryption in federated protocols like email has proved difficult, we're more likely to see the emergence of enhanced metadata protection in centralized environments with greater control.

      This is just true under the premise of a quickly moving ecosystem and does not need to be necessarily the case in general. A quickly moving ecosystem can be found in the field of social media e.g. But, on the other side, a "moving ecosystem" can also be seen in the global surveillance structures that put a pressure on developers to react quickly. This can also be seen as the competition here.

      What if the protocol is federated, but the development of the app that implements that protocol is centralized?

    4. It creates a climate of uncertainty, never knowing whether things will work or not.

      This is not a technological problem but a social one in the first place. It demands new solutions to reduce uncertainty. The automatic update mechism of Let'sEncrypt! might be a hint to what we should look at.

    5. If XMPP is so extensible, why haven't those extensions quickly brought it up to speed with the modern world?

      Is extensibility the only paradigm for updating protocols alongside the moving ecosystem? Regarding other open source tools like Wordpress the update mechanisms are more convenient (although update happen too often with WP).

    6. A recorded album can be just the same 20 years later, but software has to change.

      Concerning a physical record or tape that's correct. But if you look at the cover versions of songs of the past, it is obvious that there is the desire to reinterpret them, to hear the musical idea of a song in the contemporary cultural context. Thus, "cover protocols" are the reinterpretation and reimplementation of a protocol idea. No one would say, a cover song is an update of the original song. It is a concurrent version, a concurrent implementation of a musical idea and can be understood knowing the original or not. Certainly, music is not software, but if the code is the foundation for software, notes are the foundation for music.

  45. Nov 2016
    1. My thoughts on Climatic Associations of British Species Distributions Show Good Transferability in Time but Low Predictive Accuracy for Range Change by Rapacciuolo et al. (2012).

    2. Whilst the consensus method we used provided the best predictions under AUC assessment – seemingly confirming its potential for reducing model-based uncertainty in SDM predictions [58], [59] – its accuracy to predict changes in occupancy was lower than most single models. As a result, we advocate great care when selecting the ensemble of models from which to derive consensus predictions; as previously discussed by Araújo et al. [21], models should be chosen based on aspects of their individual performance pertinent to the research question being addressed, and not on the assumption that more models are better.

      It's interesting that the ensembles perform best overall but more poorly for predicting changes in occupancy. It seems possible that ensembling multiple methods is basically resulting in a more static prediction, i.e., something closer to a naive baseline.

    3. Finally, by assuming the non-detection of a species to indicate absence from a given grid cell, we introduced an extra level of error into our models. This error depends on the probability of false absence given imperfect detection (i.e., the probability that a species was present but remained undetected in a given grid cell [73]): the higher this probability, the higher the risk of incorrectly quantifying species-climate relationships [73].

      This will be an ongoing challenge for species distribution modeling, because most of the data appropriate for these purposes is not collected in such a way as to allow the straightforward application of standard detection probability/occupancy models. This could potentially be addressed by developing models for detection probability based on species and habitat type. These models could be built on smaller/different datasets that include the required data for estimating detectability.

    4. an average 87% of grid squares maintaining the same occupancy status; similarly, all climatic variables were also highly correlated between time periods (ρ>0.85, p<0.001 for all variables). As a result, models providing a good fit to early distribution records can be expected to return a reasonable fit to more recent records (and vice versa), regardless of whether relevant predictors of range shift have actually been captured. Previous studies have warned against taking strong model performance on calibration data to indicate high predictive accuracy to a different time period [20], [24]–[26]; our results indicate that strong model performance in a different time period, as measured by widespread metrics, may not indicate high predictive accuracy either.

      This highlights the importance of comparing forecasts to baseline predictions to determine the skill of the forecast vs. the basic stability of the pattern.

    5. Most variation in the prediction accuracy of SDMs – as measured by AUC, sensitivity, CCRstable, CCRchanged – was among species within a higher taxon, whilst the choice of modelling framework was as important a factor in explaining variation in specificity (Table 4 and Table S4). The effect of major taxonomic group on the accuracy of forecasts was relatively small.

      This suggests that it will be difficult to know if a forecast for a particular species will be good or not, unless a model is developed that can predict which species will have what forecast qualities.

    6. The correct classification rate of grid squares that remained occupied or remained unoccupied (CCRstable) was fairly high (mean±s.d.  = 0.75±0.15), and did not covary with species’ observed proportional change in range size (Figure 3B). In contrast, the CCR of grid squares whose occupancy status changed between time periods (CCRchanged) was very low overall (0.51±0.14; guessing randomly would be expected to produce a mean of 0.5), with range expansions being slightly better predicted than range contractions (0.55±0.15 and 0.48±0.12, respectively; Figure 3C).

      This is a really important result and my favorite figure in this ms. For cells that changed occupancy status (e.g., a cell that has occupied at t_1 and was unoccupied at t_2) most models had about a 50% chance of getting the change right (i.e., a coin flip).

    7. The consensus method Mn(PA) produced the highest validation AUC values (Figure 1), generating good to excellent forecasts (AUC ≥0.80) for 60% of the 1823 species modelled.

      Simple unweighted ensembles performed best in this comparison of forecasts from SDMs for 1823 species.

    8. Quantifying the temporal transferability of SDMs by comparing the agreement between model predictions and observations for the predicted period using common metrics is not a sufficient test of whether models have actually captured relevant predictors of change. A single range-wide measure of prediction accuracy conflates accurately predicting species expansions and contractions to new areas with accurately predicting large parts of the distribution that have remained unchanged in time. Thus, to assess how well SDMs capture drivers of change in species distributions, we measured the agreement between observations and model predictions of each species’ (a) geographic range size in period t2, (b) overall change in geographic range size between time periods, and (c) grid square-level changes in occupancy status between time periods.

      This is arguably the single most important point in this paper. It is equivalent to comparing forecasts to simple baseline forecasts as is typically done in weather forecasting. In weather forecasting it is typical to talk about the "skill" of the forecast, which is how much better it does than a simple baseline. In this case the the baseline is a species range that doesn't move at all. This would be equivalent to a "naive" forecast in traditional time-series analysis since we only have a single previous point in time and the baseline is simply the prediction based on this value not changing.

    9. Although it is common knowledge that some of the modelling techniques we used (e.g., CTA, SRE) generally perform less well than others [32], [33], we believe that their transferability in time is not as well-established; therefore, we decided to include them in our analysis to test the hypothesis that simpler statistical models may have higher transferability in time than more complex ones.

      The point that providing better/worse fits on held out spatial training data is not the same was providing better forecasts is important especially given the argument about simpler models having better transferability.

    10. We also considered including additional environmental predictors of ecological relevance to our models. First, although changes in land use have been identified as fundamental drivers of change for many British species [48]–[52], we were unable to account for them in our models – like most other published accounts of temporal transferability of SDMs [20], [21], [24], [25] – due to the lack of data documenting habitat use in the earlier t1 period; detailed digitised maps of land use for the whole of Britain are not available until the UK Land Cover Map in 1990 [53].

      The lack of dynamic land cover data is a challenge for most SDM and certainly for SDM validation using historical data. If would be interesting to know, in general, how much better modern SDMs become based on held out data when land cover is included.

    11. Great Britain is an island with its own separate history of environmental change; environmental drivers of distribution size and change in British populations are thus likely to differ somewhat from those of continental populations of the same species. For this reason, we only used records at the British extent to predict distribution change across Great Britain.

      This restriction to Great Britain for the model building is a meaningful limitation since Great Britain will typically represent a small fraction of the total species range for many of the species involved. However this is a common issue for SDMs and so I think it's a perfectly reasonable choice to make here given the data availability. It would be nice to see this analysis repeated using alternative data sources that cover spatial extents closer to that of the species range. This would help determine how well these results generalize to models built at larger scales.

  46. Oct 2015
    1. advocate for the preservation, archiving, and free c

      This is a really important question about the relative invisibility of e-lit. Distribution tends to rely on models which mimic academia and fandom - funny linking those two I suppose.

  47. Jan 2014
    1. Distribution of departments with respect to responsibility spheres. Ignoring the "Myself" choice, consider clustering the parties potentially responsible for curation mentioned in the survey into three "responsibility spheres": "local" (comprising lab manager, lab research staff, and department); "campus" (comprising campus library and campus IT); and "external" (comprising external data repository, external research partner, funding agency, and the UC Curation Center). Departments can then be positioned on a tri-plot of these responsibility spheres, according to the average of their respondents' answers. For example, all responses from FeministStds (Feminist Studies) were in the campus sphere, and thus it is positioned directly at that vertex. If a vertex represents a 100% share of responsibility, then the dashed line opposite a vertex represents a reduction of that share to 20%. For example, only 20% of ECE's (Electrical and Computer Engineering's) responses were in the campus sphere, while the remaining 80% of responses were evenly split between the local and external spheres, and thus it is positioned at the 20% line opposite the campus sphere and midway between the local and external spheres. Such a plot reveals that departments exhibit different characteristics with respect to curatorial responsibility, and look to different types of curation solutions.

      This section contains an interesting diagram showing the distribution of departments with respect to responsibility spheres:

      http://www.alexandria.ucsb.edu/~gjanee/dc@ucsb/survey/plots/q2.5.png