François Chollet on Twitter: "The problem with data-driven decisions: you don't see what happens to the data in the long term. What looks good for your metrics right now might kill your product in a few years. It's like AI: data gives you local generalization, but for the long term, you need principles."

Country	Code	For customers of
United States	40404	(any)
Canada	21212	(any)
United Kingdom	86444	Vodafone, Orange, 3, O2
Brazil	40404	Nextel, TIM
Haiti	40404	Digicel, Voila
Ireland	51210	Vodafone, O2
India	53000	Bharti Airtel, Videocon, Reliance
Indonesia	89887	AXIS, 3, Telkomsel, Indosat, XL Axiata
Italy	4880804	Wind
3424486444	Vodafone
» See SMS short codes for other countries

1. Adam Singer‏Verified account @AdamSinger 15 Oct 2018
  Replying to @fchollet
  
  You need analysts with deep understanding of the business, customers, roadmaps etc. They can't just sit in a silo.
  
  6 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Richard Podkolinski‏ @mrdevlar 15 Oct 2018
  Replying to @fchollet
  
  That's why data requires context. Data-driven, does not mean data-exclusive. One of the great joys of being a working statistician is you get to interact with all sorts of subject matter experts that can flush out what the data is saying.
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Giri Damerla‏ @giridamerla 15 Oct 2018
  Replying to @fchollet
  
  Yes! For long term decisions, you need principles aka vision.
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Benjamin Rouif‏ @Brouif 15 Oct 2018
  Replying to @fchollet
  
  I think it should be part of the culture. Being data driven does not mean that you fully rely on the data but use the data to challenge your ideas and you explain why you re going against the numbers ... A balance that is hard to find.
  
  1 retweet 7 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. AlgoritHuman‏ @AlgoritHuman 15 Oct 2018
  Replying to @fchollet
  
  For example, it could need 1 AI dedicated to brainstorming the data itself...while 1other AI compute data as usual, or #AlgoritHuman could manage these principles...by better coding as possible...!?
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Harshitha Machiraju‏ @nonlinear_human 15 Oct 2018
  Replying to @fchollet
  
  Problem comes you can never anticipate the new surge that comes in data...
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. trylks‏ @trylks 15 Oct 2018
  Replying to @fchollet
  
  trylks Retweeted Programming Wisdom
  
  This world be a good blogpost, elaborating on the McNamara fallacy & TD learning IMHO: data should be used to build & improve a model. The model used for decision making, & check its impact & dependencies on context for potential paradigm shifts Related:https://twitter.com/CodeWisdom/status/1046015990679719936?s=19 …
  
  trylks added,
  
  Programming Wisdom @CodeWisdom
  
  “If you can get today’s work done today, but you do it in such a way that you can’t possibly get tomorrow’s work done tomorrow, then you lose.” - Martin Fowler, Refactoring: Improving the Design of Existing Code
  
  1 like
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo