Stack Overflow 0 to 2000 in 30 Days - John Wang

---
layout: false

### Who am I?

* John Wang - aka "Grokify"
* Head of PM for RingCentral Developer Program
* Top 50 Contributor (of 900+) to OpenAPI Generator 
* Built an Alexa 12k website

* Recent focus on Product Mgmt, Product Marketing
* Weekend Warrior

* 6,785 Stack Overflow reputation - Top 0.01%
* 2,000+ reputation in 28 days of posting

---

### Head a Dev Program?

### Need a Stack Overflow Tag!

---

### 28 Days of Posts

---
### Agenda

* Why 2000?
* 7 Steps to Success
  1. Choose Identity
  2. Choose Approach
  3. Choose Tags
  4. Monitor Questions
  5. Post Answers
  6. Refine Answers
  7. Humility / General Niceness
* Results

---

# Why 2000?

---

---

### What You Get

* **edit questions and answers @ 2,000**
* *create tags @ 1,500*
* create gallery chat rooms @ 1,000
* access review queues @ 500
* view close votes @ 250
* reduce ads @ 200
* vote down @ 125
* edit community wiki, create chat rooms @ 100
* set bounties @ 75
* comment everywhere @ 50
* talk in chat  @ 20
* flag posts, vote up @ 15
* post more links, answer protected questions, create wiki posts @ 10
* participate in meta @ 5

---

## What You Really Get

---

### 7 Steps to Success

---

## Step 1

## Choose Your Identity

---

---

### Have a Secret Identity

---

### Your Identity

* Hide your identity to gain confidence

* You can reveal yourself any time
  * Users are assigned numeric ids, not usernames
  * Usernames can be changed at any time
  * Display names can be chnaged at any time

* No one cares if you are anonymous

---

## Steps 2 and 3

## Choose Your Strategy

### Reputation + Tags

---

---

### Reputation Overview

* Questions
  * 5 for upvote
  * 2 for accepting answer
* Answers
  * 15 for accept
  * 10 for upvote
  * 25-500 for bounties
* Edits
  * 2 for accepted edit

---

### Reputation Overview

* 200 point daily limit (upvotes + edits)

* Exceptions for daily limit
  * Bounties are immune to daily limit
  * Association bonus (100 points)
  * Accepting an answer (2 points)
  * Having an answer be accepted (15 points)

---

### Questions

Pros

* Don't have to be an expert

Cons

* Need to post questions that haven't been asked
* Need to ask questions that are "on-topic"
* Esoteric questions may not get much traffic

---

### Answers

Pros

* Lots of questions to answer
* More points: 10/upvote, 15/acceptance, bounties
* Builds your expertise (for hiring)

Cons

* Have to be fast for simple questions

---

### Edits

Pros

* Simple
* Little expertise needed
* Can just fix project name

Cons

* Doesn't build expertise
* Caps out at 2,000 reputation

---

### Why Answers?

* Builds expertise
* Good for Developer Relations
* Less frustration vs. questions
* Higher velocity vs. edits

---

### How Many Answers?

* Estimate how many answers you need
* Easy way: go to Users tab and sort by Reputation

```
SELECT
  owner_user_id AS user_id,
  COUNT(id) AS answer_count,
  CONCAT(
    "https://stackoverflow.com/users/",
    CAST(owner_user_id AS STRING)) AS user_url
FROM `bigquery-public-data.stackoverflow.posts_answers`
WHERE owner_user_id IN (
  SELECT id FROM `bigquery-public-data.stackoverflow.users`
  WHERE reputation=2000
  ORDER BY creation_date DESC
  LIMIT 100 )
GROUP BY owner_user_id
ORDER BY answer_count DESC
```

---

### Answer Counts

* Less than 200, estimate 100

---

## Step 3

## Choose Tags

---

### Focus: Choosing Tags

* Build expertise
* Accelerate ability to post good answers

* Need flow of new questions
* Healthy asker : answerer ratio

---

### Question Velocity by Tag

* Need new questions to answer
* Google BigQuery with public dataset

```sql
SELECT
  tag,
  COUNT(tag) AS count_last_3_months,
  CAST(COUNT(tag) / 90 AS INT64) AS count_per_day,
  CAST(COUNT(tag) / 90 / 24 AS INT64) AS count_per_hour
FROM (
  SELECT SPLIT(q.tags, '|') AS tag
  FROM `bigquery-public-data.stackoverflow.posts_questions` q
  WHERE DATETIME(q.creation_date) >= (
    SELECT DATETIME_SUB(DATETIME(creation_date), INTERVAL 3 MONTH)
    FROM `bigquery-public-data.stackoverflow.posts_questions`
    ORDER BY creation_date DESC
    LIMIT 1
  )
), UNNEST(tag) AS tag
GROUP BY tag
ORDER BY count_last_3_months DESC
```

---

### Question Velocity by Tag

---

### My Experience

* Answer any questions I can
* Currently use Go and Ruby
* Know some JavaScript, Python, PHP from years ago

```sql
SELECT
 tag,
 COUNT(tag) AS vote_and_accept_count
FROM (
 SELECT SPLIT(q.tags, '|') AS tag
 FROM `bigquery-public-data.stackoverflow.votes` v
 LEFT JOIN `bigquery-public-data.stackoverflow.posts_answers` a
 ON v.post_id = a.id
 LEFT JOIN `bigquery-public-data.stackoverflow.posts_questions` q
 ON a.parent_id = q.id
 WHERE a.owner_user_id = 1908967
 AND v.vote_type_id IN (1, 2)
 AND DATE(v.creation_date) < DATE('2015-06-21')
), UNNEST(tag) AS tag
GROUP BY tag
ORDER BY vote_and_accept_count DESC
LIMIT 10
```

---

### My Tag Vote Counts

---

### My Tag "Quality"

---

## Step 4

## Monitor Questions

---

### General Approach

1. Answer anything you can
2. Bounties with few answers
3. Any topic you want to learn more about

---

### Question Classification

* Maximize your contribution

* New questions have short *half-life*
  * Easy *for you* questions are worth answering first
  * Answer others before traffic decay grows
* Bounties stay active for up to 7 days

---

## Steps 5 & 6

### Post Answers
### Refine Answers

---

### Answer Techniques

1. Answer should work
  * If code, try to verify answer before publishing
    * Non-working answer is worse than no answer
  * If reference, link to official docs
2. Show working code
  * Optionally use Stack Snippets, JSFiddle, regex101, Go Playground
3. Use references
  * Links, quotes, images
  * More official is better: MDN, IETF, Google, etc.

* For easy questions
  * Step 1: Provide a minimal, correct answer ASAP
  * Step 2: Refine answer to provide more info

---

### Simple Example - Question

* Answer ASAP, fully, with authority

---

### Simple Example - Answer

* Answered 1 minute after accepted answer
* Explain the issue
* Provide a specific solution
* Provide a generial example
* Link to authoritative references
* 4 answers. Scores: 15, **10**, 0

---

### Simple Reference - Question

* Documentation question, but popular

---

### Simple Reference - Answer

* Answer the question
* Provide an authoritative quote
* Link to official documentation
* 4 answers. Scores: **85**, 10, 10, 0

---

### Regular Example - Question

* Reasonable question (not just RTFD)

---

### Regular Example - Answer

* Provide working solution
* Proveable solution via JSFiddle
* 1 answer. Score: **85**

---

### Regular Example - JSFiddle

* JSFiddle is nice
* Doesn't work on a ship

---

### Bounty Example - Question

* Bounties get more views
* 505 views

---

### Bounty Example - Answer

* Organize answer into 2 answers
* Proveable examples via Stack Snippets
* 2 answers. Score: **160**, 10*

---

### Popular, Late Example - Question

* Researching for yourself
* SO shows up in search
* Accepted and multiple answers
* Doesn't answer the question for you

---

### Popular, Late - Answer

* Answered over 2.5 years later
* 5 answers. Scores: 1335, 940, **710**, 150, 50

---

### Be Tenacious

---

## Step 7

## Have a Good Attitude

---

### Be Humble and Generally Nice

* Be polite
* Be helpful
* You are here make others successful
* Don't worry
  * Can be anonymous
  * Can delete answers and questions

---

## My Results
### Reputation: 2,115
### Posting Days: 28
### Answer Count: 99
### Question Count: 6
### Bounty Points: 100

---

### Calendar Date: 2015-06-21

* Get date using Reputation API
* Built Go SDK using OpenAPI Spec

```golang
import(
	stackoverflow "github.com/grokify/go-stackoverflow/client"
	"github.com/grokify/go-stackoverflow/util"
)

func main() {
  apiClient := stackoverflow.NewAPIClient(stackoverflow.NewConfiguration())
  history, err := util.GetReputationHistoryAll(
    apiClient, "stackoverflow", "1908967")
  if err != nil {
    log.Fatal(err)
  }
  dayRep, err := history.DateForReputation(int32(2000))
  if err != nil {
    log.Fatal(err)
  }
  fmt.Printf("DATE [%v]\n", dayRep.Day)
}
```

Code: https://github.com/grokify/stackoverflow-the-hard-way/blob/main/examples/reputation/main.go

---

### Post Days Query

* Use date in query

```
WITH post_dates AS (
 SELECT DISTINCT(post_date)
 FROM (
 (
 SELECT DATE(creation_date) AS post_date
 FROM `bigquery-public-data.stackoverflow.posts_questions`
 WHERE owner_user_id = 1908967
 AND DATE(creation_date) < DATE('2015-06-21')
 GROUP BY post_date
 )
 UNION ALL
 (
 SELECT DATE(creation_date) AS post_date
 FROM `bigquery-public-data.stackoverflow.posts_answers`
 WHERE owner_user_id = 1908967
 AND DATE(creation_date) < DATE('2015-06-21')
 GROUP BY post_date
 )
 )
 GROUP BY post_date
)
SELECT COUNT(post_date) AS post_days_count
FROM post_dates
```

---

### Post Days Results

---

## Parting Thoughts

#### Be Comfortable
#### Be Focused
#### Be Tenacious
#### Be Humble
#### Be Helpful

---

### Thank you

Contact & Code: John Wang, aka Grokify

* https://twitter.com/grokify
* https://github.com/grokify
* https://stackoverflow.com/users/1908967/grokify
* https://github.com/grokify/stackoverflow-bigquery

Credits

* [Stack Overflow API](https://stackoverflow.com/) - Data
* [Google BigQuery](https://bigquery.cloud.google.com/) - Analytics
* [OpenAPI Generator](https://github.com/OpenAPITools/openapi-generator) - SDK Tools
* [draw.io](https://www.draw.io/) - SVG Tools
* [Remark.js](https://remarkjs.com/) - Presentation Tools