Thread by @hondanhon

Things that can all be true:

1) last-mile, vaccine software can be really complicated when taking into account the 50 states and 3,006 counties in the U.S.

2) procurement regulations can be biased toward incumbents who have previously done business

3) previous business with government is not in itself an indicator of previous successful outcomes nor a predictor in itself of future successful outcomes

4) $44m may actually be a reasonable development cost, all considered

5) previous specific domain experience (e.g. last-mile vaccine distribution) is not itself an indicator or predictor of future success

6) success increases with the degree of executive sponsorship, clear mandate and day-to-day involvement of executives

7) the talent, skills and experience for delivering last-mile vaccine distribution likely exists in government at federal and state level

8) jurisdictions/government entities from individual departments to counties to states can believe they all have unique requirements

9) they also in practice have shared requirements, and that proportion can increase if policy and practice are directed to be flexible

10) government software infrastructure has much to learn from the private sector, including the private consumer sector

11) the private consumer technology sector has the luxury of being able to ignore or not meet user needs that government must meet

12) procurement rules or practice can be biased toward “transfer” systems, preferring selection of an existing system with customization

13) the usage of transfer systems do not necessarily reduce risk or improve outcomes

14) it is too hard to transfer “good” systems

15) on close reading, many government regulations, at all levels, do not actually prevent the delivery or development user-centered, iterative, interoperable software

16) many in government may believe that regulations prevent (15) above

17) this is not a technology problem

18) this not a technology problem, it’s actually program and policy problem

19) this is not a program and policy problem, it’s actually an operations problem

20) this is not an operations problem, it’s actually a procurement problem

21) government lacks the skills ability to develop and operate critical software so it must be outsourced to the private sector

22) private sector outsourcing is as much about risk management (“a single throat to choke”) as it is about ensuring successful outcomes

23) volunteers and the public can play a critical role in last-mile vaccine distribution

24) volunteers and the public need to understand how government works in practice to better and more productively engage

25) government needs to understand how to better harness volunteers

26) volunteer software infrastructure fills a critical current need

27) volunteer software infrastructure contributes to learned helplessness

28) negotiating and more importantly prioritizing the requirements of up to 3,006 counties is a difficult job that also requires actually making prioritization decisions

29) fax machines are actually a critical part of healthcare infrastructure in the united states

30) nobody really wants or would choose for fax machines to be a critical part of healthcare infrastructure in the united states

31) many people who might have to use a last-mile vaccine distribution system may not be “expert” computer users

32) many people who might have to use such a system may be experts in the systems they currently use

33) it is not necessarily the job of a specific “technology” procurement to effectively make up for potential shortfalls in computer literacy

34) technology systems will be used by those who are not computer literate

35) related to (29) and (30) the continued usage of fax machines can be a sign of how fax machines meet a user need that must be met or exceeded in any replacement or alternative systems

36) “we can’t tell people how to do their jobs, or that they have to use this new system”

37) you can and do tell people how to do their jobs, and that they have to use a new system

38) in many cases, the underlying mainframe systems and their databases or effective systems-of-record for programs are reliable, stable and able to meet surges in demand

39) the software on top of underlying mainframe systems (e.g. web access) are frequently inflexible

40) increasing the effectiveness, efficiency, accessibility (and more) of a program depends on timely metrics, telemetry, instrumentation

41) metrics require someone to make a decision about targets, which is not a technology problem

42) metrics are probably going to make someone look bad

43) most people don’t want to look bad and will try to avoid it, which might include opposing metrics and targets in general

44) “innovation” is stifled in government by hard rules, regulations and legislation

45) “innovation” is stifled in government by inaccurate and uncountered beliefs

46) “innovation” actually does happen in government, and government is capable of it

47) re government learning from private industry and vice versa:

https://twitter.com/harrisj/status/1355656030869463040

48) time and materials contracts are actually a reasonable way to pay for effective software development

49) deliverables-based contracts are actually a reasonable way to pay for things that involve the use of technology

50) it is too hard, or it is impossible, or it is prohibited to use or procure certain modern software development techniques, tools or platforms

51) modern software development techniques, tools or platforms are not prohibited and are available to use

(I’m going to stop for now)

More resources/further reading:

1. A Civic Technologist’s Practice Guide by @cydharrell at cydharrell.com/book/

2. This is how America gets it vaccines by Cat Ferguson and Karen Hao at technologyreview.com/2021/01/27/101…

3. The 18F Guide to De-Rising Technology Projects at derisking-guide.18f.gov

4. “When can I schedule a COVID-19 vaccine appointment?” Why the government can’t answer this simple question by @hanaschank and @raskarmit at fastcompany.com/90593617/when-…

5. “On Design Thinking”, by Maggie Gram in n+1 magazine: nplusonemag.com/issue-35/revie…

6. This thread by my former @codeforamerica Summit co-chair @corizarek:

https://twitter.com/corizarek/status/1353847909444247555?s=20

7. The prompt for this thread, “What went wrong with America’s $44 million vaccine data system?” by @biocuriosity

technologyreview.com/2021/01/30/101…