Meeting Notes 2015-01-12 (TODC)
編輯歷史
| 時間 | 作者 | 版本 |
|---|---|---|
| 2015-11-16 08:43 – 08:44 | r878 – r904 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-09-05 12:18 | r877 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-09-05 07:01 | r876 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-02-04 20:41 – 20:53 | r853 – r875 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-23 20:12 – 20:13 | r848 – r852 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-13 10:22 – 10:24 | r804 – r847 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-13 08:42 – 08:42 | r802 – r803 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-13 07:32 – 07:40 | r737 – r801 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-13 05:45 | r736 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 15:56 – 17:20 | r730 – r735 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 15:45 – 15:45 | r728 – r729 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 14:51 – 14:51 | r726 – r727 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 14:46 – 14:47 | r707 – r725 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 14:31 – 14:31 | r690 – r706 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:50 – 13:58 | r574 – r689 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:50 – 13:50 | r571 – r573 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:49 – 13:49 | r567 – r570 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:48 | r566 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:44 – 13:46 | r562 – r565 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:28 | r561 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 13:07 – 13:17 | r539 – r560 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 12:50 – 12:50 | r536 – r538 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 12:27 – 12:28 | r526 – r535 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 12:26 – 12:26 | r524 – r525 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 11:28 – 12:24 | r391 – r523 | |
顯示 diff（diff 過大，略過）
|
||
| 2015-01-12 11:28 | r390 | |
顯示 diff（272 行未修改）
|
||
| 2015-01-12 11:00 – 11:28 | r383 – r389 | |
顯示 diff（20 行未修改）
tonyq> I’m working in Jaclyn’s office as consultant as of this week.
- pofeng> Current process is that gov.tw cherry-picks data into structural open data, so if there is a central archive, there is no implicit exclusivity. All we need is a well-known API that other participants can join.
+ pofeng> Current泵 process is that gov.tw cherry-picks data into structural open data, so if there is a central archive, there is no implicit exclusivity. All we need is a well-known API that other participants can join.
au> I can help designing the API, no problem.
pofeng> I have a few concerns though:
（242 行未修改）
*All user-and-operator visible products of this efforts are to be OSI-compatible open sourced and developed in the open.
*User-visible processes, if patentable, needs to be granted non-exclusively in the usual fashion.
+
+
+ 第
|
||
| 2015-01-12 09:57 | r382 | |
顯示 diff（262 行未修改）
*yhsiang mentioned would love to lead the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/145 )
- *wildsky mentioned would help the frontend part ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/151 )
+ *wildsky mentioned would help the front-end part ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/151 )
*suensummit mentioned would love to in the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/164 )
*Licensing
（2 行未修改）
|
||
| 2015-01-12 09:55 | r381 | |
顯示 diff（263 行未修改）
*yhsiang mentioned would love to lead the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/145 )
*wildsky mentioned would help the frontend part ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/151 )
- *suensummit mentioned would love to in the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/164 )
+ *suensummit mentioned would love to in the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/164 )
*Licensing
*All user-and-operator visible products of this efforts are to be OSI-compatible open sourced and developed in the open.
（1 行未修改）
|
||
| 2015-01-12 09:42 – 09:54 | r296 – r380 | |
顯示 diff（261 行未修改）
+ *yhsiang mentioned would love to lead the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/145 )
+ *wildsky mentioned would help the frontend part ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/151 )
+ *suensummit mentioned would love to in the Dev team ( http://logbot.g0v.tw/channel/g0v.tw/2015-01-12/164 )
*Licensing
*All user-and-operator visible products of this efforts are to be OSI-compatible open sourced and developed in the open.
（1 行未修改）
|
||
| 2015-01-12 09:28 – 09:30 | r279 – r295 | |
顯示 diff（3 行未修改）
Location: Hope Bay Technology, Neihu, Taipei
Participants: pofeng, whisky, ben, au, tonyq
- Tentative architecture (subject to revision):
+ T
+ entative architecture (subject to revision):
*
- Preamble
+ P
+ Second revision, per NDC's feedback:
+
+ *
+ reamble
ben> (off record) briefing about initiation and prepratory efforts on gathering the initial fund of this effort.
（249 行未修改）
|
||
| 2015-01-12 09:04 – 09:12 | r276 – r278 | |
顯示 diff（52 行未修改）
ben> yeah but that will be in 2016.
- whisky> yes so for 2015, you can market this effort as a prototypical pilot, but in 2016 there’s no prevention for NDC and other agencies and run through ZTHY’s portals. currently it’s a meta-data read-only mindset.
+ whisky> yes so for 2015, you can marakehis effort as a prototypical pilot, but in 2016 there’s no prevention for NDC and other agencies and run through ZTHY’s portals. currently it’s a meta-data read-only mindset.
ben> if ZTHY claimed this work but cannot deliver this, I won’t give a damn.
（139 行未修改）
ben> no, just OpenStack stuff.
- whisky> google has acquhired an UN data visualization team.
+ whisky> google has acquhied an UN data visualization team.
tonyq> our expectation of gov is rather different — ben and I would like to work from inside to get all data out, but whisky would like to see them as a partner.
（60 行未修改）
|
||
| 2015-01-12 08:45 – 09:01 | r207 – r275 | |
顯示 diff（30 行未修改）
pofeng> as long as the output is open-source then it is worth it.
- ben> NDC will push data in well-known format into an archival house; it’s just we establish a new prototype for it. Foundational effort takes care of the data center, incuding rental of VM (standard OpenStack hosting fees, with educational and NGO discouts).
+ ben> NDC will push data in well-known format into an archival house; it’s just we establish a new prototype for it. Foundational effort takes care of the data center, incuding rental of VM (standard OpenStack hosting fees, with educational and NGO discoutsn).
ben> unstructured data is fine, but we of course prefer structural data. the data structuraling effort performed here can also get its own credit.
（222 行未修改）
*Licensing
- *All user-and-operator visible products of this efforts are to be OSI-c
+ *All user-and-operator visible products of this efforts are to be OSI-compatible open sourced and developed in the open.
+ *User-visible processes, if patentable, needs to be granted non-exclusively in the usual fashion.
|
||
| 2015-01-12 08:45 | r206 | |
顯示 diff（260 行未修改）
|
||
| 2015-01-12 08:23 – 08:45 | r144 – r205 | |
顯示 diff Meeting Notes 2015-01-12
-
+ This document is CC BY 4.
+ ttps://g0v.hackpad.com/R4w7Sd5Ti46
Location: Hope Bay Technology, Neihu, Taipei
Participants: pofeng, whisky, ben, au, tonyq
（5 行未修改）
ben> (off record) briefing about initiation and prepratory efforts on gathering the initial fund of this effort.
- au> demo: g0vxgov training, vtaiwan.tw, hackathon in gov(now shifted to off-site facilities), data.g0v.tw ng
- ben> (off record) debriefing of Simon and Terry’s previous discussions.
+ au> demo: g0vxgov training, vtaiwan.tw, hackathon in gov(n ow shifted to off-site facilities), dadata.g0v.tw ngen> (off record) debriefing of Simon and Terry’s previous discussions.
Discussions
（67 行未修改）
pofeng> we can do training course on structural data. how quick do you want commitment?
- ben> we want to tell NDC “there is a foundation for this”. of course funding won’t arrive. but why chairperson Pofeng votes **against** this proposal first?
+ ben> we want to tell NDC “there is a foundation for this”. of course funding won’t arrive immediately. but why chairperson Pofeng voted against this proposal first?
pofeng> because we lack the staff power for it. this is originally our plan for 2016, and 2015 is really too early.
（63 行未修改）
ben> by March, data.gov.tw’s metadata needs to appear on archive.tw. then the filtering engine’s process needs to be up on public deliberation.
- tonyq> catalog is first priority, as well as most of data does not filtering.
+ tonyq> catalog is first priority, as well as most of data that does not require filtering.
tonyq> for kptaipei, the unofficial website actually has very few data set in term of quantity, but the process and quality is more of a useful PR value. this is more important than raw quantity.
（99 行未修改）
*Ben is in charge of overall direction
*Implementation needs PM and Legal and Dev personnel
+
+
+ *Licensing
+ *All user-and-operator visible products of this efforts are to be OSI-c
|
||
| 2015-01-12 08:10 | r143 | |
顯示 diff（256 行未修改）
|
||
| 2015-01-12 07:55 – 08:05 | r1 – r142 | |
顯示 diff- Untitled
+ Meeting Notes 2015-01-12
- This pad text is synchronized as you type, so that everyone viewing this page sees the same text. This allows you to collaborate seamlessly on documents!
+ Location: Hope Bay Technology, Neihu, Taipei
+ Participants: pofeng, whisky, ben, au, tonyq
+ Tentative architecture (subject to revision):
+
+ *
+
+ Preamble
+
+ ben> (off record) briefing about initiation and prepratory efforts on gathering the initial fund of this effort.
+ au> demo: g0vxgov training, vtaiwan.tw, hackathon in gov(now shifted to off-site facilities), data.g0v.tw ng
+ ben> (off record) debriefing of Simon and Terry’s previous discussions.
+
+ Discussions
+
+ tonyq> I’m working in Jaclyn’s office as consultant as of this week.
+ pofeng> Current process is that gov.tw cherry-picks data into structural open data, so if there is a central archive, there is no implicit exclusivity. All we need is a well-known API that other participants can join.
+ au> I can help designing the API, no problem.
+ pofeng> I have a few concerns though:
+ *OCF.tw needs auditing and financial effort to get more than 10M in funding
+ *What is the timeline? If more than two month then a new foundation is not likely possible.
+ *For Tax purposes, we need to spend more than 70% of funding, so if there is a new foundation next year then we need to sign a preparatory contract for fund transfer.
+
+ ben> the sponsor’s consensus is that I’m the trusted manager of this NTD$5M-per-sponsor fund.
+
+ pofeng> the difference between donating-for-credit versus selling-for-credit; for tax purposes it is different.
+
+ ben> selling-for-credit of access is fine.
+
+ pofeng> as long as the output is open-source then it is worth it.
+
+ ben> NDC will push data in well-known format into an archival house; it’s just we establish a new prototype for it. Foundational effort takes care of the data center, incuding rental of VM (standard OpenStack hosting fees, with educational and NGO discouts).
+
+ ben> unstructured data is fine, but we of course prefer structural data. the data structuraling effort performed here can also get its own credit.
+
+ pofeng> the API interface, the receiver, needs to be open source.
+
+ ben> NDC will be pushing, we do not need to pull.
+
+ pofeng> as long as all involved software stack as well as web it’s OSI-compliant, I’m OK about it.
+
+ ben> we just need a well-known FTP-like metadata and data convention for NDC to push into it. we need a full-text search engine, with standard warranty and disclaimer.
+
+ ben> as for publishing and working on government data, how the participants get credit, needs a structural design.
+
+ whisky> NDC will open up the API; they just want to build a catalog instead of maintaining its own mirror. however all local governments start to host their own data, instead of pushing — there’s no easy way to get exclusivity.
+
+ ben> per-5-minute pushing, can NDC take up this much bandwidth host?
+
+ whisky> they are already prepared to absorb the cost.
+
+ ben> yeah but that will be in 2016.
+
+ whisky> yes so for 2015, you can market this effort as a prototypical pilot, but in 2016 there’s no prevention for NDC and other agencies and run through ZTHY’s portals. currently it’s a meta-data read-only mindset.
+
+ ben> if ZTHY claimed this work but cannot deliver this, I won’t give a damn.
+
+ whisky> I bring up metadata because NDC’s metadata is very unpolished; the data cleaning up service is very much the most effort. NDC’s own data may have 3-star quality, but the main effort will be in cleaning up in local agencies.
+
+ au> getting officials to use metadata-aware process is crucial.
+
+ pofeng> how is NDC’s gathering process going on?
+
+ ben> NDC’s folks will be here shortly and we can check with him.
+
+ ben> we are not talking about agencies publishing the API, but raw Database and Tape Archival of all EY agencies. Local agencies may choose to use the same archival process, but of course we can only suggest.
+
+ tonyq> so EY is a showcase for local governments.
+
+ ben> it’s a race between Mao and Ko actually.
+
+ whisky> filtering process?
+
+ ben> technical metadata: all DB SQL schema is to be published first. privacy and secret-level fields are to be deliberated, but by NDC in a public forum instead of the Archive.
+
+ au> so anything passed off the firewall is already clear of identifiable information?
+
+ ben> yes.
+
+ tonyq> actually, each agency’s SQL schema differ wildly. so this platform is a good way for participants to contribute converters.
+
+ pofeng> we can do training course on structural data. how quick do you want commitment?
+
+ ben> we want to tell NDC “there is a foundation for this”. of course funding won’t arrive. but why chairperson Pofeng votes **against** this proposal first?
+
+ pofeng> because we lack the staff power for it. this is originally our plan for 2016, and 2015 is really too early.
+
+ ben> the location of OCF needs to be in Taipei.
+
+ pofeng> it’s Bureau of Culture’s Taipei City foundation.
+
+ au> CL and NDC had previous discussions, but today I cannot speak for CL, just architectural references.
+
+ au> (demo of archive.org structure.)
+
+ ben> archive.tw Foundation?
+
+ au> OK. (registered)
+
+ pofeng> any recommendation on accountants?
+
+ ben> no worries; I’ll sort this out.
+
+ pofeng> …and we need full-timers for this.
+
+ ben> at least 2-3 full timers are needed.
+
+ pofeng> for 2015, if this is in OCF’s task force, then would 3-5 representatives be too much?
+
+ ben> tonyq, can you get off-hook from Jaclyn’s efforts to work on this instead? or at least recommendations?
+
+ au> ronny has http://data.g0v.ronny.tw/ and may be interested in technical consulting, if not full-timer.
+
+ whisky> OpenFoundry has a team already versed in open data hosting.
+
+ ben> how large should the team be? I think 10 is not too much. the funding is plenty.
+
+ whisky> also muyueh with https://sheetdb.com/veggie.muyueh.com/ has previous experiences on presentation and delivery of semi-structured data in structured fashion.
+
+ pofeng> we just need a head for this effort, for both Tech and PM. I’ll get contact with PM candidates.
+
+ ben> what’s the fallback?
+
+ pofeng> whisky?
+
+ whisky> (uncommitted)
+
+ tonyq> I would like to work with NDC and agencies as the gathering and structural consultant.
+
+ pofeng> OK we have a volunteer!
+
+ ben> we can start with already-open data from data.gov.tw and data.g0v.tw as the pilot, and work on privacy and de-identification later.
+
+ tonyq> what is risks?
+
+ ben> just our reputations.
+
+ pofeng> I worry about my clinic’s tax records… (lol) anyway the Steering Committee is my idea of reducing political risk.
+
+ tonyq> I don’t really care about my career this year or anything.
+
+ ben> This is the right time. Mao has buy-in, and I actually helped Simon… (off record)… so it’s very likely to succeed. the data catalog as brought up by tony will be resolved and is critical.
+
+ pofeng> the largest risk still lies in the execution.
+
+ ben> Folks NDC has a more open mindset as they are not through the normal career-ladder.
+
+ whisky> deliverable?
+
+ ben> by March, data.gov.tw’s metadata needs to appear on archive.tw. then the filtering engine’s process needs to be up on public deliberation.
+
+ tonyq> catalog is first priority, as well as most of data does not filtering.
+
+ tonyq> for kptaipei, the unofficial website actually has very few data set in term of quantity, but the process and quality is more of a useful PR value. this is more important than raw quantity.
+
+ ben> media strategy is another matter.
+
+ whisky> NDC’s update interval?
+
+ au> (demo) 3du’s consensus with terms.naer.edu.tw on exporting in sql.gz format.
+
+ whisky> can we collect from edu.tw directly instead of going through ndc?
+
+ ben> sure, as long as they use the same endpoints as NDC, agencies can bypass NDC’s cataloging pipeline and write directly into the filtering process.
+
+ au> confirmed? then we are changing the diagram here.
+
+ pofeng> medical cloud and national palace museum are somewhat counter-example here, which raised TAHR’s attention, even ended up in lawsuit. so this is why “release for research purpose” term is added. I feel somewhat responsible for this and will work with medical workers on this purpose.
+
+ au> yeah, one-way functions on national ID really need to be used as one-time uuids if one wants complete anonymization.
+
+ ben> but then we don’t have foreign keys anymore.
+
+ au> that’s TAHR’s main position actually.
+
+ pofeng> anyway their current proposal is double-hashing. this is very contentious. so we’re leaning on opt-in.
+
+ au> opt-out policies needs a full deliberation process before committing on anything. this is part of why we’re working on vtaiwan.tw.
+
+ pofeng> arranging 1:1 meeting time with PM candidates.
+
+ ben> (checking schedule) I’m fully booked during daytime
+
+ au> perhaps asynchronous telecommunication?
+
+ ben> OK. here’s my phone number and contact details…
+
+ pofeng> shall we call ronny now?
+
+ au> maybe after PM commits? ronny has PoCs ready and so we can work out details once the system flow is determined.
+
+ whisky> can we work on application package (sandboxed apps running on archive.tw) here?
+
+ au> is ben’s team working on Kubernetes support?
+
+ ben> no, just OpenStack stuff.
+
+ whisky> google has acquhired an UN data visualization team.
+
+ tonyq> our expectation of gov is rather different — ben and I would like to work from inside to get all data out, but whisky would like to see them as a partner.
+
+ ben> my expectation of gov is just for them not to block citizen’s process.
+
+ pofeng> so OpenFoundry actually closes on end of 2015 not 2016, but it’s still OK to talk with PM candidates.
+
+ pofeng> e.g. http://drugs.olc.tw/ has received data retraction requests, so our catalog’s revision needs to be fast enough that pulls it down in case of emergency.
+
+ au> among Ben’s 8 ideas on FB page, there is only one that is most impractical: “3b. 立法明定：只要有適當之錯誤回報及修正機制，政府人員對過去資料之錯誤不用負行政責任” because damage caused by administrative data is still judiciary’s issue.
+
+ pofeng> maybe if they provide it for free, one can sign a waiver to not sue government?
+
+ au, whisky> that does not preclude damage.
+
+ ben> actually I explicitly said “行政責任” as waiving administrative responsibility, not 刑事 or 民事...
+
+ whisky> yeah but that does cause a chilling effect when brought up (e.g. TAHR’s case).
+
+ whisky> open data in the current quality cannot be value-added, it needs massive cleaning up and structuring efforts.
+
+ ben> so this is a challenge. the archive needs to provide forked/cleaned-up revisions from the public.
+
+ whisky> so gov is no longer the only “canonical” source in this view, as we need to establish a culture that says errors is normal.
+
+ whisky> so we need a 行政命令.
+
+ pofeng> 函釋 doesn’t need to bring to LY and may be even faster.
+
+ tonyq> ben’s plan calls on agencies to unconditionally backup, this is prone to encounter resistance.
+
+ au> but that is ben’s plan’s key point. my point is that EY agencies are not using metadata-aware tools to produce content, and so if you forcibly collect data, all you will get is unstructured data.
+
+ tonyq> e.g. in taipei.gov.tw, they say “we can open all data but we don’t know the meaning or schema of this”. so my position here is about removing obstacle.
+
+ au> this calls for CfA-style ambassadors into agencies if we need structured data.
+
+ pofeng> tonyq, are you really willing to amass this team of ambassadors?
+
+ tonyq> this is Jacylin’s plan for me actually, as an IT fellow for the entire EY. I’m not exactly sure I’m fit for this role yet — it’s a rather heavy burden.
+
+ au> already suggested unstructured-data checking fields to NDC.
+
+ whisky> how to pull back patches and forked versions in a structural way to NDC?
+
+ au> directly read-back to agencies can’t be done in a forced way, but we can provide support materials.
+
+ ben> updating diagram to include contributing-back policies.
+
+ Action Items
+
+ *HR — Personnel (pofeng)
+ *PM candidates
+ *posting on g0v hackpad (au - done)
+ *Funding and Structure (ben)
+ *Ben is in charge of overall direction
+ *Implementation needs PM and Legal and Dev personnel
|
||
| 2015-01-12 07:55 | r0 | |
顯示 diff+ Untitled
+ This pad text is synchronized as you type, so that everyone viewing this page sees the same text. This allows you to collaborate seamlessly on documents!
|
||