{
  "id": 177377,
  "title": "Licenses - CC-BY-NC-SA-4.0",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/177377",
  "author_name": "",
  "post_date": "2020-08-25T18:00:58.158171100Z",
  "votes": 6,
  "comment_count": 4,
  "views": 0,
  "content": "<p>This challenge looks very interesting, with what looks to be a lot of time and care taken to prep the data. I noticed the license required of the solutions is CC-BY-NC-SA-4.0. </p>\n<p>I don't have much experience with this specific license so I wanted to get this conversation started early since licensing concerns have been a sore point in some other recent Kaggle challenges.</p>\n<p>After taking a quick look it would seem like </p>\n<ul>\n<li>Many strong-copyleft licenses like GPL based licenses (GPL-2, GPL-3) are not compatible with CC-BY-NC-SA-4.0 </li>\n<li>Weak copyleft like MPL should be okay</li>\n<li>Permissive licenses like Apache 2.0, BSD, MIT are okay as long as their reqs for attribution, non-advertising, etc are met and they don't have any specific clauses conflicting the CC-BY-NC-SA-4.0</li>\n</ul>\n<p>Would the above be a correct summary?</p>",
  "messages": [
    {
      "id": "985419",
      "postDate": "08/25/2020 18:00:58",
      "content": "<p>This challenge looks very interesting, with what looks to be a lot of time and care taken to prep the data. I noticed the license required of the solutions is CC-BY-NC-SA-4.0. </p>\n<p>I don't have much experience with this specific license so I wanted to get this conversation started early since licensing concerns have been a sore point in some other recent Kaggle challenges.</p>\n<p>After taking a quick look it would seem like </p>\n<ul>\n<li>Many strong-copyleft licenses like GPL based licenses (GPL-2, GPL-3) are not compatible with CC-BY-NC-SA-4.0 </li>\n<li>Weak copyleft like MPL should be okay</li>\n<li>Permissive licenses like Apache 2.0, BSD, MIT are okay as long as their reqs for attribution, non-advertising, etc are met and they don't have any specific clauses conflicting the CC-BY-NC-SA-4.0</li>\n</ul>\n<p>Would the above be a correct summary?</p>",
      "rawMarkdown": "This challenge looks very interesting, with what looks to be a lot of time and care taken to prep the data. I noticed the license required of the solutions is CC-BY-NC-SA-4.0. \n\nI don't have much experience with this specific license so I wanted to get this conversation started early since licensing concerns have been a sore point in some other recent Kaggle challenges.\n\nAfter taking a quick look it would seem like \n* Many strong-copyleft licenses like GPL based licenses (GPL-2, GPL-3) are not compatible with CC-BY-NC-SA-4.0 \n* Weak copyleft like MPL should be okay\n* Permissive licenses like Apache 2.0, BSD, MIT are okay as long as their reqs for attribution, non-advertising, etc are met and they don't have any specific clauses conflicting the CC-BY-NC-SA-4.0\n\nWould the above be a correct summary?",
      "votes": null
    },
    {
      "id": "985528",
      "postDate": "08/25/2020 19:34:18",
      "content": "<blockquote>\n  <p>The answer to your question largely depends on </p>\n  <p>1) the external open-source software (OSS) that you plan to use<br>\n  2) the license type and its specific terms<br>\n  3) and how you use the OSS software with the competition materials.  </p>\n  <p>The CC-BY-NC-SA 4.0 license allows sharing and adapting the materials as long as you provide attribution, use the materials for noncommercial purposes, and distribute your contributions under the same license.  So you are correct that you can run into compatibility issues with using GPL-2.0 (for example) because GPL-2.0 does not allow you to place further restrictions on that software.  Since CC-BY-NC-SA 4.0 would place the noncommercial use restriction on the GPL-2.0 code, that would probably result in incompatibility.</p>\n  <p>Also, you might run into compatibility issues with certain weak copyleft code as well.  For example, MPL 2.0 includes the language under 3.1 Distribution of Source Form: “You may not attempt to alter or restrict the recipients’ rights in the Source Code Form.”  So by placing the noncommercial use restriction noted above on the MPL 2.0 code, you might see an incompatibility issue there.  Again, largely depends on a number of factors described here.  Permissive should generally be okay if the license does not prohibit restrictions and we comply with the terms for attribution (as you noted)</p>\n</blockquote>\n<p>Does this help?</p>",
      "rawMarkdown": ">  The answer to your question largely depends on \n\n> 1) the external open-source software (OSS) that you plan to use\n> 2) the license type and its specific terms\n> 3) and how you use the OSS software with the competition materials.  \n\n> The CC-BY-NC-SA 4.0 license allows sharing and adapting the materials as long as you provide attribution, use the materials for noncommercial purposes, and distribute your contributions under the same license.  So you are correct that you can run into compatibility issues with using GPL-2.0 (for example) because GPL-2.0 does not allow you to place further restrictions on that software.  Since CC-BY-NC-SA 4.0 would place the noncommercial use restriction on the GPL-2.0 code, that would probably result in incompatibility.\n\n> Also, you might run into compatibility issues with certain weak copyleft code as well.  For example, MPL 2.0 includes the language under 3.1 Distribution of Source Form: “You may not attempt to alter or restrict the recipients’ rights in the Source Code Form.”  So by placing the noncommercial use restriction noted above on the MPL 2.0 code, you might see an incompatibility issue there.  Again, largely depends on a number of factors described here.  Permissive should generally be okay if the license does not prohibit restrictions and we comply with the terms for attribution (as you noted)\n\nDoes this help?",
      "votes": null
    },
    {
      "id": "985538",
      "postDate": "08/25/2020 19:45:31",
      "content": "<p>I guess I jumped ahead a bit there. In terms of 'compatibility' the issue of concern for most is what restrictions the challenge code licenses impose on licenses of 3rd party Python modules, libraries, other imported code (github, etc) that a contestant might include in their end solution. </p>\n<p>For example, in the recent Wheat challenge (object detection) a lot of contestants had to make a last minute switch away from Yolov5 because its licensed GPL-3 and the challenge required code to be released as MIT.</p>",
      "rawMarkdown": "I guess I jumped ahead a bit there. In terms of 'compatibility' the issue of concern for most is what restrictions the challenge code licenses impose on licenses of 3rd party Python modules, libraries, other imported code (github, etc) that a contestant might include in their end solution. \n\nFor example, in the recent Wheat challenge (object detection) a lot of contestants had to make a last minute switch away from Yolov5 because its licensed GPL-3 and the challenge required code to be released as MIT.",
      "votes": null
    },
    {
      "id": "985568",
      "postDate": "08/25/2020 20:26:53",
      "content": "<p>The story is similar here. Yolov5 or any other library with GPL-3 is not allowed.</p>",
      "rawMarkdown": "The story is similar here. Yolov5 or any other library with GPL-3 is not allowed.",
      "votes": null
    },
    {
      "id": "986531",
      "postDate": "08/26/2020 15:17:37",
      "content": "<p>I wish I had participated in the survey you did earlier, but it looks like a good feature perhaps more so for Kaggle to develop itself than for a host to work on, would be an option in the competition overview where the host can specify their own license requirement(s) and then a matrix of common ✅ compatible and ⛔️ incompatible licenses are shown. This would help the lay people like myself efficiently sort through frameworks and models for possible inclusion in our solutions. Maybe it doesn't matter much for this competition since you're doing all the heavy lifting of manually inspecting the lawyer-speak licenses belonging to each package someone proposes, but most hosts aren't as embedded into the community as you are.</p>",
      "rawMarkdown": "I wish I had participated in the survey you did earlier, but it looks like a good feature perhaps more so for Kaggle to develop itself than for a host to work on, would be an option in the competition overview where the host can specify their own license requirement(s) and then a matrix of common ✅ compatible and ⛔️ incompatible licenses are shown. This would help the lay people like myself efficiently sort through frameworks and models for possible inclusion in our solutions. Maybe it doesn't matter much for this competition since you're doing all the heavy lifting of manually inspecting the lawyer-speak licenses belonging to each package someone proposes, but most hosts aren't as embedded into the community as you are.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 985528,
      "author_name": "iglovikov",
      "author_url": "",
      "post_date": "08/25/2020 19:34:18",
      "content": "<blockquote>\n  <p>The answer to your question largely depends on </p>\n  <p>1) the external open-source software (OSS) that you plan to use<br>\n  2) the license type and its specific terms<br>\n  3) and how you use the OSS software with the competition materials.  </p>\n  <p>The CC-BY-NC-SA 4.0 license allows sharing and adapting the materials as long as you provide attribution, use the materials for noncommercial purposes, and distribute your contributions under the same license.  So you are correct that you can run into compatibility issues with using GPL-2.0 (for example) because GPL-2.0 does not allow you to place further restrictions on that software.  Since CC-BY-NC-SA 4.0 would place the noncommercial use restriction on the GPL-2.0 code, that would probably result in incompatibility.</p>\n  <p>Also, you might run into compatibility issues with certain weak copyleft code as well.  For example, MPL 2.0 includes the language under 3.1 Distribution of Source Form: “You may not attempt to alter or restrict the recipients’ rights in the Source Code Form.”  So by placing the noncommercial use restriction noted above on the MPL 2.0 code, you might see an incompatibility issue there.  Again, largely depends on a number of factors described here.  Permissive should generally be okay if the license does not prohibit restrictions and we comply with the terms for attribution (as you noted)</p>\n</blockquote>\n<p>Does this help?</p>",
      "votes": null,
      "replies": [
        {
          "id": 985538,
          "author_name": "rwightman",
          "author_url": "",
          "post_date": "08/25/2020 19:45:31",
          "content": "<p>I guess I jumped ahead a bit there. In terms of 'compatibility' the issue of concern for most is what restrictions the challenge code licenses impose on licenses of 3rd party Python modules, libraries, other imported code (github, etc) that a contestant might include in their end solution. </p>\n<p>For example, in the recent Wheat challenge (object detection) a lot of contestants had to make a last minute switch away from Yolov5 because its licensed GPL-3 and the challenge required code to be released as MIT.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 985568,
          "author_name": "iglovikov",
          "author_url": "",
          "post_date": "08/25/2020 20:26:53",
          "content": "<p>The story is similar here. Yolov5 or any other library with GPL-3 is not allowed.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 986531,
          "author_name": "authman",
          "author_url": "",
          "post_date": "08/26/2020 15:17:37",
          "content": "<p>I wish I had participated in the survey you did earlier, but it looks like a good feature perhaps more so for Kaggle to develop itself than for a host to work on, would be an option in the competition overview where the host can specify their own license requirement(s) and then a matrix of common ✅ compatible and ⛔️ incompatible licenses are shown. This would help the lay people like myself efficiently sort through frameworks and models for possible inclusion in our solutions. Maybe it doesn't matter much for this competition since you're doing all the heavy lifting of manually inspecting the lawyer-speak licenses belonging to each package someone proposes, but most hosts aren't as embedded into the community as you are.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "985419": "This challenge looks very interesting, with what looks to be a lot of time and care taken to prep the data. I noticed the license required of the solutions is CC-BY-NC-SA-4.0. \n\nI don't have much experience with this specific license so I wanted to get this conversation started early since licensing concerns have been a sore point in some other recent Kaggle challenges.\n\nAfter taking a quick look it would seem like \n* Many strong-copyleft licenses like GPL based licenses (GPL-2, GPL-3) are not compatible with CC-BY-NC-SA-4.0 \n* Weak copyleft like MPL should be okay\n* Permissive licenses like Apache 2.0, BSD, MIT are okay as long as their reqs for attribution, non-advertising, etc are met and they don't have any specific clauses conflicting the CC-BY-NC-SA-4.0\n\nWould the above be a correct summary?",
    "985528": ">  The answer to your question largely depends on \n\n> 1) the external open-source software (OSS) that you plan to use\n> 2) the license type and its specific terms\n> 3) and how you use the OSS software with the competition materials.  \n\n> The CC-BY-NC-SA 4.0 license allows sharing and adapting the materials as long as you provide attribution, use the materials for noncommercial purposes, and distribute your contributions under the same license.  So you are correct that you can run into compatibility issues with using GPL-2.0 (for example) because GPL-2.0 does not allow you to place further restrictions on that software.  Since CC-BY-NC-SA 4.0 would place the noncommercial use restriction on the GPL-2.0 code, that would probably result in incompatibility.\n\n> Also, you might run into compatibility issues with certain weak copyleft code as well.  For example, MPL 2.0 includes the language under 3.1 Distribution of Source Form: “You may not attempt to alter or restrict the recipients’ rights in the Source Code Form.”  So by placing the noncommercial use restriction noted above on the MPL 2.0 code, you might see an incompatibility issue there.  Again, largely depends on a number of factors described here.  Permissive should generally be okay if the license does not prohibit restrictions and we comply with the terms for attribution (as you noted)\n\nDoes this help?",
    "985538": "I guess I jumped ahead a bit there. In terms of 'compatibility' the issue of concern for most is what restrictions the challenge code licenses impose on licenses of 3rd party Python modules, libraries, other imported code (github, etc) that a contestant might include in their end solution. \n\nFor example, in the recent Wheat challenge (object detection) a lot of contestants had to make a last minute switch away from Yolov5 because its licensed GPL-3 and the challenge required code to be released as MIT.",
    "985568": "The story is similar here. Yolov5 or any other library with GPL-3 is not allowed.",
    "986531": "I wish I had participated in the survey you did earlier, but it looks like a good feature perhaps more so for Kaggle to develop itself than for a host to work on, would be an option in the competition overview where the host can specify their own license requirement(s) and then a matrix of common ✅ compatible and ⛔️ incompatible licenses are shown. This would help the lay people like myself efficiently sort through frameworks and models for possible inclusion in our solutions. Maybe it doesn't matter much for this competition since you're doing all the heavy lifting of manually inspecting the lawyer-speak licenses belonging to each package someone proposes, but most hosts aren't as embedded into the community as you are."
  },
  "source": "meta"
}