Cole, Lee
Writings on Open Gestalts and the Origins of Ideaflow: IdeaOverflow Paper & Notes
gestaltexplanation.jacobcole.net
Prof. Marvin Minsky
6.868 Society of Mind
IdeaOverflow IDE (Idea Development Environment) Idea Matching Algorithm
The basis for a platform for collaborative ideation and project sharing.
Jacob Cole, Holden Lee
A MindMap of this paper is available at http://app.wisemapping.com/c/maps/124943/public . The matching algorithm is available at https://github.com/tmad4000/IdeaOverflow_som and the web platform source code is available at https://github.com/tmad4000/ideaoverflow-hackny . A demo of the web app is available at http://pacific-tundra-3629.herokuapp.com/ or (old) http://agile-ocean-3244.herokuapp.com/.
While the Internet currently has many tools to facilitate discussion of people and events, there is no centralized platform optimized to facilitate deep, collaborative thinking.Thus we started IdeaOverflow Idea Development Environment (IDE), an online system that facilitates collaborative cognition by, as a key feature, finding relationships between ideas inputted into a database.
First, we contend that IdeaOverflow is a relevant, modern frame for the problem of strong artificial intelligence that refines the antiquated Turing test, and makes actionable Douglas Hofstadter’s theory that analogy is the core of cognition.
Second, we develop and apply theories of how humans make connections between ideas and use existing tools (MIT ConceptNet) to take the first steps towards implementing an idea-connection machine. We built a basic web platform, implemented two rudimentary algorithms that find conceptually related ideas, and we lay a road map forward. Even at this early stage, our algorithm often successfully makes relevant connections between ideas drawn from hackathonprojects.tk, a public, crowdsourced database of over 100 startup company/hackathon project ideas.
Finally, in Section 3, we apply gestalt theory in a novel way to understand how individuals come up with ideas and apply this understanding to project the process of ideation into a shared space. We demonstrate how the platform as a whole can be used to create new good ideas.
The great difficulty in pushing ideation into a shared space isthat people can only make connections between the ideas in their own minds. The ability to make these connections is a fundamental aspect of human intelligence. Talking to a friend working on a certain idea, a person can suggest related ideas and people that the collaborator should investigate. However, currently, when those ideas are stored in an external database rather than a single person’s mind, this ability to make connections is lost; static databases are not “intelligent”. As knowledge becomes increasingly distributed, we need to make connections between ideas that are not held in the mind of a single person, or even the minds within a personal network.
We propose the following task as an alternative to the Turing Test: program an AI to emulate humans’ ability to find ideas related to a given idea. Given a database of ideas and an idea within the database, the program should find the top 10 to 20 ideas related to that idea, so that the resulting list is indistinguishable from that compiled by a team of humans who have spent extensive time evaluating the list for connections. Like a human collaborator, a system that could do this would suggest related ideas and people that a user should investigate. In this way, it could have a genuinely constructive and intelligent "conversation" with its users about the ideas they are working on. Better than any individual human collaborator, it would try to find the top 10 to 20 ideas out of the database of possibly millions, whereas a human can only search on a much more limited data set.
Read-only v0. To edit an item, click its content to open the focused page. Column resize, sort, inline edit, and the per-list item_schema editor are tracked under ticket 1z3.
| — |
| inherit |
| — |
| 7 | Figure 2.1-1: Sample of IdeaOverflow database entries | — | — | — | inherit | — |
| 9 | Figure 2.1-2: Finding related ideas by keyword search | — | — | — | inherit | — |
| 13 | I. Matching by related concepts | — | — | — | inherit | — |
| 15 | II. Matching by problem/solution | — | — | — | inherit | — |
| 16 | We can refine matching by related concepts if we first understand the components of an idea. An idea has two main components to it. | — | — | — | inherit | — |
| 18 | - A method of solution, or technology that it utilizes. For instance, does it use crowdsourcing to accomplish something that would take an inordinate amount of time otherwise? | — | — | — | inherit | — |
| 21 | III. Abstraction Hypothesis: | — | — | — | inherit | — |
| 23 | IV. Matching by Transitivity | — | — | — | inherit | — |
| 28 | V. Story Hypothesis: | — | — | — | inherit | — |
| 32 | Figure 2.3-1. Command line interface for matching ideas | — | — | — | inherit | — |
| 33 | Linking ideas using commonsense knowledge | — | — | — | inherit | — |
| 34 | We used MIT ConceptNet 4 and Divisi2 to write one implementation of the method how_related_are_ideas(idea1, idea2), which works as follows. | — | — | — | inherit | — |
| 35 | - Extract a list of concepts from each idea. | — | — | — | inherit | — |
| 36 | - For each concept in the first list, find the most related idea in the second list using ConceptNet and Divisi. | — | — | — | inherit | — |
| 37 | - Average all the relatedness scores. | — | — | — | inherit | — |
| 39 | Transitivity | — | — | — | inherit | — |
| 42 | Figure 2.4-1: Ideas most related to cooking app query | — | — | — | inherit | — |
| 43 | It also lists the most related concepts between the two ideas along with the relatedness score. | — | — | — | inherit | — |
| 44 | Figure 2.4-2. Most related concepts between cooking app query and GrabLunchWith app | — | — | — | inherit | — |
| 50 | - Combine the commonsense framework with a story understanding engine such as Genesis. | — | — | — | inherit | — |
| 51 | There would also be value in creating a battery of standard tests researchers could run their idea relation algorithms against, to concretely measure their performance. | — | — | — | inherit | — |
| 53 | Figure 2.5-1. Mockup of clustered graph visualization of ideas | — | — | — | inherit | — |
| 56 | Figure 3.1-1. Kanisza triangle illusion | — | — | — | inherit | — |
| 59 | ## | — | — | — | inherit | — |
| 61 | - First we are hungry. | — | — | — | inherit | — |
| 62 | - Thus we want to eat, and one common pathway of resolving this is cooking. | — | — | — | inherit | — |
| 63 | - Cooking requires first deciding what to cook, and then obtaining ingredients. | — | — | — | inherit | — |
| 66 | - First, it would see that “deciding what to cook” is a subgoal of “cook.” | — | — | — | inherit | — |
| 71 | Barlow, H. (1961) "Possible principles underlying the transformation of sensory messages" in Sensory Communication, MIT Press | — | — | — | inherit | — |
| 74 | Hutter M. Universal Algorithmic Intelligence: A mathematical top->down approach. Artificial General Intelligence, 2007, Springer, pages 227-290. | — | — | — | inherit | — |
| 75 | Minsky, Marvin. 1986. The Society of Mind. Simon & Schuster, Inc., New York, NY | — | — | — | inherit | — |
| 76 | [Salomon, Gavriel (1997). Distributed cognitions: Psychological and educational considerations](http://books.google.com/books?id=m8Yna0cjxAgC). Cambridge University Press. | — | — | — | inherit | — |
| 77 | Gentner, Dedre, Keith James Holyoak, and Boicho N. Kokinov. The Analogical Mind: | — | — | — | inherit | — |
| 78 | Perspectives from Cognitive Science. Cambridge, MA: MIT, 2001. Print. 499. | — | — | — | inherit | — |
| 80 | # | — | — | — | inherit | — |
| 82 | On Open Gestalts https://psycho-therapeut.ch/EN/gestalttherapie-e.htm | psycho-therapeut.ch | — | — | inherit | — |
| 83 | - “An open "Gestalt" could be for example an unresolved conflict, an unfulfilled wish or an unprocessed traumatic experience.” | — | — | — | inherit | — |
| 84 | On Gestalt Psychology https://en.wikipedia.org/wiki/Gestalt_psychology | en.wikipedia.org | — | — | inherit | — |
| 85 | - The background in a piece of art is the closed gestalt: the attention skates over it | — | — | — | inherit | — |
| 86 | - The foreground is an open gestalt: that which tugs on your attention | — | — | — | inherit | — |
| 87 | - Attentional selection can change foreground and background as in Faces & Vases Illusion. Similarly, backgrounded observations can connect in our minds to tug on our attention. | — | — | — | inherit | — |
| 88 | Process the open gestalts in societies mind → heal the society mind. | — | — | — | inherit | — |
| 89 | Further directions to explore | — | — | — | inherit | — |
| 90 | - Buddhist concept of “emptiness” and all phenomena as gestalt phenomena | — | — | — | inherit | — |
| 91 | - Gestalts and their relation to Liebniz’ Monads | — | — | — | inherit | — |
| 92 | See also: http://ideaflowbackground.jacobcole.net/ | ideaflowbackground.jacobcole.net | — | — | inherit | — |
| 93 | Attached is the core logic idea_mapper.py of the idea matching algorithm and the content of the database of 29 ideas, idea_db.txt. | — | — | — | inherit | — |
| 94 | [omitted] | — | — | — | inherit | — |
| 95 | idea_db.txt | — | — | — | inherit | — |
| 96 | TITLE: Workflow sharing | — | — | — | inherit | — |
| 97 | DESCRIPTION: Take the workflows of top students and make or sell complete sets of customized materials to help other people adopt the workflows of the super productive people. | — | — | — | inherit | — |
| 99 | PROBLEM: People aren’t productive. | — | — | — | inherit | — |
| 100 | SOLUTION: Share workflows of productive people. | — | — | — | inherit | — |
| 101 | TECHNOLOGY: | — | — | — | inherit | — |
| 102 | TITLE: MedTime | — | — | — | inherit | — |
| 103 | DESCRIPTION: Track when people actually take their meds | — | — | — | inherit | — |
| 104 | MORE: what if you used an app to track when people actually /do/ take their meds | — | — | — | inherit | — |
| 105 | this is absolutely what we should be thinking about | — | — | — | inherit | — |
| 106 | there exist a number of drugs for which the main and important factor is the minimum blood level concentration is the important part | — | — | — | inherit | — |
| 107 | among these are birth control, seizure meds, beta-blockers and a whole bunch of other things | — | — | — | inherit | — |
| 108 | and you have to take them at the same time every day because you need to maintain minimum effective concentration | — | — | — | inherit | — |
| 109 | but, excepting beta-blockers, having a concentration in excess, up to about 3x the normal does, is not something that will hurt you | — | — | — | inherit | — |
| 110 | so, what would be cooler than reminding you to take your meds when you tend to sit down and eat dinner | — | — | — | inherit | — |
| 111 | is an app that analyzes when you actually do take your meds, and then suggests custom dosages based on your schedule | — | — | — | inherit | — |
| 113 | it's not something that's anywhere outside of the realm of physicians to prescribe, in fact, its actually in their best interest | — | — | — | inherit | — |
| 114 | PROBLEM: People don’t take their meds on schedule. | — | — | — | inherit | — |
| 115 | SOLUTION: Track when people actually take their meds, and remind them. | — | — | — | inherit | — |
| 116 | TECHNOLOGY: | — | — | — | inherit | — |
| 117 | TITLE: Pitch-based scrolling | — | — | — | inherit | — |
| 118 | DESCRIPTION: Use voice to scroll. | — | — | — | inherit | — |
| 119 | MORE: Pitch-based scrolling - e.g. when you sing middle C, the document jumps to 50% mark. natural tool for voice recognition users/RSI/paralyzed patients | — | — | — | inherit | — |
| 120 | PROBLEM: People who are paralyzed can’t scroll. | — | — | — | inherit | — |
| 121 | SOLUTION: Use pitch to scroll. | — | — | — | inherit | — |
| 122 | TECHNOLOGY: | — | — | — | inherit | — |
| 123 | TITLE: EEG-based Slumping Detector | — | — | — | inherit | — |
| 124 | DESCRIPTION: Monitor posture and help correct it. | — | — | — | inherit | — |
| 126 | PROBLEM: Slumping, bad posture. | — | — | — | inherit | — |
| 127 | SOLUTION: Monitor posture and help correct it. | — | — | — | inherit | — |
| 128 | TECHNOLOGY: EEG | — | — | — | inherit | — |
| 129 | TITLE: WorldThroughYourEyes | — | — | — | inherit | — |
| 130 | DESCRIPTION: Track where a person’s eyes are focusing, and analyze the effects on their lives. | — | — | — | inherit | — |
| 132 | PROBLEM: People want to see the world differently. | — | — | — | inherit | — |
| 133 | SOLUTION: Track where a person’s eyes are focusing. | — | — | — | inherit | — |
| 134 | TECHNOLOGY: Google Glasses | — | — | — | inherit | — |
| 135 | TITLE: InstaBoxSite | — | — | — | inherit | — |
| 136 | DESCRIPTION: Build an easy to modify "suggestion box" framework anyone can use to quickly build a website off that mold. | — | — | — | inherit | — |
| 139 | PROBLEM: Many websites use the same kind of suggestion box. | — | — | — | inherit | — |
| 140 | SOLUTION: Make a framework for suggestion boxes. | — | — | — | inherit | — |
| 141 | TECHNOLOGY: | — | — | — | inherit | — |
| 142 | TITLE: virtual reality game with 3d-goggles in swimming pool. | — | — | — | inherit | — |
| 143 | DESCRIPTION: virtual reality game with 3d-goggles in swimming pool or indoor skydiving arena | — | — | — | inherit | — |
| 145 | PROBLEM: | — | — | — | inherit | — |
| 146 | SOLUTION: | — | — | — | inherit | — |
| 147 | TECHNOLOGY: 3-D goggles | — | — | — | inherit | — |
| 148 | TITLE: Index card sharing | — | — | — | inherit | — |
| 149 | DESCRIPTION: Organize and share your thoughts by writing them concisely and linking them together. | — | — | — | inherit | — |
| 153 | There are two parts to this project: allowing the user to create a system of cards, and then facilitating sharing. | — | — | — | inherit | — |
| 155 | PROBLEM: People’s thoughts are messy. | — | — | — | inherit | — |
| 156 | SOLUTION: Write them down on virtual index cards and organize them. | — | — | — | inherit | — |
| 157 | TECHNOLOGY: | — | — | — | inherit | — |
| 158 | TITLE: Thoughtstream tool | — | — | — | inherit | — |
| 159 | DESCRIPTION: Make a central platform for thoughtstreaming. | — | — | — | inherit | — |
| 161 | PROBLEM: People have ideas that they don’t record. | — | — | — | inherit | — |
| 162 | SOLUTION: Make a central platform for thoughtstreaming. | — | — | — | inherit | — |
| 163 | TECHNOLOGY: | — | — | — | inherit | — |
| 164 | TITLE: Flow Me | — | — | — | inherit | — |
| 165 | DESCRIPTION: Suggest rhymes and alliterations as you type. | — | — | — | inherit | — |
| 167 | PROBLEM: People writing poems can’t think of rhymes. | — | — | — | inherit | — |
| 168 | SOLUTION: Suggest rhymes and alliterations as you type. | — | — | — | inherit | — |
| 169 | TECHNOLOGY: | — | — | — | inherit | — |
| 170 | TITLE: DeviceSymphony | — | — | — | inherit | — |
| 171 | DESCRIPTION: a mobile or Web app that plays the same note you are singing | — | — | — | inherit | — |
| 173 | PROBLEM: | — | — | — | inherit | — |
| 174 | SOLUTION: | — | — | — | inherit | — |
| 175 | TECHNOLOGY: | — | — | — | inherit | — |
| 176 | TITLE: Fridgey | — | — | — | inherit | — |
| 177 | DESCRIPTION: Gives recipes based on what you have in your fridge. | — | — | — | inherit | — |
| 179 | Comments: | — | — | — | inherit | — |
| 183 | http://www.instantfundas.com/2010/08/15-websites-to-find-recipes-by.html <- it exists. | — | — | — | inherit | — |
| 184 | PROBLEM: You have a bunch of stuff in the fridge, but nothing to eat. | — | — | — | inherit | — |
| 185 | SOLUTION: Gives recipes based on what you have in your fridge. | — | — | — | inherit | — |
| 186 | TECHNOLOGY: | — | — | — | inherit | — |
| 187 | TITLE: Build eyetracking heatmap from laptop webcam | — | — | — | inherit | — |
| 188 | DESCRIPTION: Build eyetracking heatmap from laptop webcam to evaluate how you read websites. | — | — | — | inherit | — |
| 190 | PROBLEM: | — | — | — | inherit | — |
| 191 | SOLUTION: | — | — | — | inherit | — |
| 192 | TECHNOLOGY: eyetracking, webcam | — | — | — | inherit | — |
| 193 | TITLE: Google docs optimized for instant web publishing | — | — | — | inherit | — |
| 194 | DESCRIPTION: Google docs optimized for instant web publishing | — | — | — | inherit | — |
| 196 | ssign a google doc to a .tk domain | — | — | — | inherit | — |
| 197 | PROBLEM: | — | — | — | inherit | — |
| 198 | SOLUTION: | — | — | — | inherit | — |
| 199 | TECHNOLOGY: | — | — | — | inherit | — |
| 200 | TITLE: Google doc tools | — | — | — | inherit | — |
| 201 | DESCRIPTION: Track traffic on Google Docs and quickly assign a google doc to a .tk domain. | — | — | — | inherit | — |
| 202 | MORE: Way to track traffic on Google Docs/quick way to assign a google doc to a .tk domain #<->Google docs optimized for instant web publishing ~jcole@mit.edu | — | — | — | inherit | — |
| 203 | PROBLEM: | — | — | — | inherit | — |
| 204 | SOLUTION: | — | — | — | inherit | — |
| 205 | TECHNOLOGY: | — | — | — | inherit | — |
| 206 | TITLE: Emergency Pizza Button app | — | — | — | inherit | — |
| 207 | DESCRIPTION: One-click pizza ordering via a big red button | — | — | — | inherit | — |
| 209 | PROBLEM: Ordering pizza takes time. | — | — | — | inherit | — |
| 210 | SOLUTION: One-click pizza ordering | — | — | — | inherit | — |
| 211 | TECHNOLOGY: | — | — | — | inherit | — |
| 212 | TITLE: PhotoQuote | — | — | — | inherit | — |
| 213 | DESCRIPTION: Photograph a quote, find the quote online, and record it. | — | — | — | inherit | — |
| 215 | PROBLEM: | — | — | — | inherit | — |
| 216 | SOLUTION: | — | — | — | inherit | — |
| 217 | TECHNOLOGY: OCR | — | — | — | inherit | — |
| 218 | TITLE: Instaquote | — | — | — | inherit | — |
| 219 | DESCRIPTION: Autocomplete quotes, create a database of quotes you’re interested in. | — | — | — | inherit | — |
| 220 | MORE: Instaquote -- site that autocompletes quotes like instadefine.com and it basically ends up letting you create a database of quotes you're interested in | — | — | — | inherit | — |
| 222 | PROBLEM: | — | — | — | inherit | — |
| 223 | SOLUTION: | — | — | — | inherit | — |
| 224 | TECHNOLOGY: | — | — | — | inherit | — |
| 225 | TITLE: FoodMom | — | — | — | inherit | — |
| 226 | DESCRIPTION: App that tells you what to cook, automatically orders ingredients (or makes you a shopping list), and suggests who you should cook with. | — | — | — | inherit | — |
| 228 | PROBLEM: | — | — | — | inherit | — |
| 229 | SOLUTION: | — | — | — | inherit | — |
| 230 | TECHNOLOGY: | — | — | — | inherit | — |
| 231 | TITLE: Call for Fall | — | — | — | inherit | — |
| 232 | DESCRIPTION: Detect when an old person has fallen, and makes an emergency call. | — | — | — | inherit | — |
| 234 | PROBLEM: Old people fall. | — | — | — | inherit | — |
| 235 | SOLUTION: Detect when an old person has fallen, and makes an emergency call. | — | — | — | inherit | — |
| 236 | TECHNOLOGY: accelerometer | — | — | — | inherit | — |
| 237 | TITLE: ThisIsHowYouSoundRightNow | — | — | — | inherit | — |
| 238 | DESCRIPTION: Determines your alertness level by using machine learning on keystrokes | — | — | — | inherit | — |
| 240 | PROBLEM: We can’t tell if we’re being productive. | — | — | — | inherit | — |
| 241 | SOLUTION: Determines your alertness level by using machine learning on keystrokes | — | — | — | inherit | — |
| 242 | TECHNOLOGY: Machine learning | — | — | — | inherit | — |
| 243 | TITLE: News Timelines | — | — | — | inherit | — |
| 244 | DESCRIPTION: Organize related news articles into timelines so you can see the meaning of each article | — | — | — | inherit | — |
| 245 | MORE: Organize related news articles into timelines so you can see the meaning of each article, and maybe join in! #intentiondb ~jcole@mit.edu | — | — | — | inherit | — |
| 246 | PROBLEM: It’s hard to follow news. | — | — | — | inherit | — |
| 247 | SOLUTION: Organize related news articles into timelines | — | — | — | inherit | — |
| 248 | TECHNOLOGY: | — | — | — | inherit | — |
| 249 | TITLE: CuriosityThread | — | — | — | inherit | — |
| 250 | DESCRIPTION: Show the chain of questions or interests a person asked or followed to gain the knowledge and passion they have. | — | — | — | inherit | — |
| 252 | PROBLEM: Follow someone’s learning process. | — | — | — | inherit | — |
| 253 | SOLUTION: Record and show the chain of questions or interests a person asked or followed to gain the knowledge and passion they have. | — | — | — | inherit | — |
| 254 | TECHNOLOGY: | — | — | — | inherit | — |
| 255 | TITLE: IdeaOverflow | — | — | — | inherit | — |
| 256 | DESCRIPTION: Allow people to share good ideas that they don't necessarily have time to implement themselves online. | — | — | — | inherit | — |
| 258 | PROBLEM: Ideas are not equally distributed in people. | — | — | — | inherit | — |
| 259 | SOLUTION: Allow people to share good ideas that they don't necessarily have time to implement themselves. | — | — | — | inherit | — |
| 260 | TECHNOLOGY: | — | — | — | inherit | — |
| 261 | TITLE: PoliticalProgressBar | — | — | — | inherit | — |
| 266 | PROBLEM: Voters can’t gauge how successful government is. | — | — | — | inherit | — |
| 267 | SOLUTION: A simple GUI that voters can use to quickly see what government is trying to accomplish and how far it has gotten towards achieving its goals. | — | — | — | inherit | — |
| 268 | TECHNOLOGY: | — | — | — | inherit | — |
| 269 | TITLE: EveryDayOfYourLife | — | — | — | inherit | — |
| 270 | DESCRIPTION: Ask a user to answer daily questions for self-reflection and share answers | — | — | — | inherit | — |
| 271 | MORE: | — | — | — | inherit | — |
| 272 | PROBLEM: | — | — | — | inherit | — |
| 273 | SOLUTION: | — | — | — | inherit | — |
| 274 | TECHNOLOGY: | — | — | — | inherit | — |
| 275 | TITLE: Minimalist homepages | — | — | — | inherit | — |
| 276 | DESCRIPTION: Quick workflow to make minimalist homepage for yourself using Google docs and tk domains | — | — | — | inherit | — |
| 277 | MORE: | — | — | — | inherit | — |
| 278 | PROBLEM: | — | — | — | inherit | — |
| 279 | SOLUTION: | — | — | — | inherit | — |
| 280 | TECHNOLOGY: | — | — | — | inherit | — |
| 281 | TITLE: GrabLunchWith app | — | — | — | inherit | — |
| 282 | DESCRIPTION: Suggests where you should have lunch, and recommends friends to go to the same place so they meet. | — | — | — | inherit | — |
| 283 | MORE: | — | — | — | inherit | — |
| 284 | PROBLEM: | — | — | — | inherit | — |
| 285 | SOLUTION: | — | — | — | inherit | — |
| 286 | TECHNOLOGY: | — | — | — | inherit | — |
| 287 | TITLE: Academic conversations tool | — | — | — | inherit | — |
| 288 | DESCRIPTION: matches people that have mentioned titles to the same Wikipedia articles in their chats, or quotes | — | — | — | inherit | — |
| 289 | MORE: | — | — | — | inherit | — |
| 290 | PROBLEM: | — | — | — | inherit | — |
| 291 | SOLUTION: | — | — | — | inherit | — |
| 292 | TECHNOLOGY: | — | — | — | inherit | — |
| 293 | 4 | — | — | — | inherit | — |