Social Blockchain

What is the value in this case of using a blockchain?

6yr ago

Social Blockchain

Maker

@jeroendeprest The main value of storing the activity in an immutable, cryptographically verifiable transaction log is 1. to keep a reliable permanent history of the data that anybody could use in the future, and 2. to be able to verify the validity and accuracy of the analytics we generate for the content consumption service. We're generating a new block roughly once every hour. Then we generate insights and try to find interesting traits about that data as a snapshot of social activity around the world at that time. This time frequency gives a nice hourly, historical view of the data. It's fun to be able to jump back in time to a specific date at a specific time and see what was trending.

6yr ago

@bryant_williams How do you handle false data?

6yr ago

@bryant_williams are those points actual problems?

6yr ago

@bryant_williams I am always quite sceptical of blockchain projects because at my job we have a lot of people coming to us with project that don't even need blockchain.

6yr ago

Simeon

@bryant_williams @jeroendeprest This would be ok as a specific feature set but not an actual product and then marketing what should be invisible (blockchain) to the average consumer.

6yr ago

Elena Zhizhimontova

This is curious! Which blockchain do you guys use? Which social activity do you collect atm (from which sources I guess)?

6yr ago

James Drummond

@zelena Yes the idea and implementation are great but blockchains usually have technical limitations (i.e slow lookups) and cost money. How does this work from a technical / business point of view?

6yr ago

Social Blockchain

Maker

Hi @zelena ! We're using a homegrown solution. The blockchain node logic is quite simple at this point because we're only running a single node. Each transaction (a social sharing event) is stored in our own custom ledger database (storing raw historical data in S3 and storing metadata, hashes, and insights in MongoDB). It was very interesting to see AWS release their own ledger database recently. Currently the primary data source is publishers using ShareThis Sharing Buttons (https://www.sharethis.com/platfo...) which is roughly 3 million publishers. Publishers using AddThis, SumoMe, or AddToAny can also contribute their site's sharing activity to the ledger by installing the socialblockchain plugin: https://www.socialblockchain.co/... .

6yr ago

Social Blockchain

Maker

@zelena @desertedisland Great point James. Many blockchains suffer from being slow and expensive. We're avoiding the cost issue right now by using a single node on a private blockchain. Each hour we inspect and analyze the recent transactions (sharing activity), then we generate analytics and insights from that and store those in MongoDB indexed by topic, category, domain, and date. So the front-end at socialblockchain.co and the data pulled is all cacheable via CDN and quick to lookup in the database. In the future, we hope to give users control to view their entire sharing history. In that case we'll have to start indexing the raw data by the user's identifier which will increase our costs slightly, but it still won't be a technical blocker.

6yr ago

Colin Winhall

TallyUp

I was surprised just how interesting I found this project, it's really well made, congrats guys! I feel that there is a lot of value held in this data that is underutilized, do you have any plans to release an API at any point?

6yr ago

Social Blockchain

Maker

@colinwinhall Thanks Colin! We agree that there's a lot of value still left to be uncovered in the data. We would definitely be open to releasing an API to help other builders. Do you have any ideas about what you might do with the API or any particular content you'd want to use? Right now all stories are primarily indexed by domain, topic, category, and time. We'll get it on our roadmap to work on an API others can use :)

6yr ago

Colin Winhall

TallyUp

@bryant_williams Hi Bryant, I'm particularly interested in exploring the data in various ways. One way would be in terms of NLP for the title of the article and the body to uncover any specific patterns that might make an article more "shareable".