I'm planning on doing a more in-depth write up on how NER/NLP in the tweetosphere is different that general NER/NLP which will by me technical explanation of what I did here. That should be up sometime later this week.
Can you elaborate more on how having these specific tweets helps your specific startup or how you are converting them from casual tweeters to your new users?
Right now there is a big disconnect in the post. I am guessing its because you assume the reader understands even what ner/nlp is
Originally the post was going to be a bit more technical but I decided to keep is simple and see what the reaction was to it. I'll make sure to put that in the technical post. For now I've included an image that I hope gives readers a bit more of an understanding on how we're using this application.
I put the blog post on Ghost droplet with Digital Ocean in a hurry as my first attempt with using Medium never ended up on the front page. I'll add one as quick as I can. For now follow me on twitter @trevorstarick. :)
Check out our API for your NER needs, we cover quite a large corpus of people, places & things. Might be able to broaden your scope: https://www.repustate.com/semantic/
Quite a bit but that might be because of our market. We're a P2P travel / sharing economy metasearch engine and the demographic we aim to target seems to enjoy traveling without burning a hole in their wallets. I'd like it to be a higher but its passive growth hacking with almost a $0 customer acquisition cost.
Have you considered what potential there is for scaling in up? Because if all you are getting that way is 100 uniques a day, the effort involved sounds like it'd make it a net loss compared to simply buying the traffic via paid advertising. If the tweets continue to generate additional referrals down the line, then of course it might change.
But frankly, having seen the travel space up close, I find this system more interesting than your startup...
It's something I've thought about and both our current investors and potential future investors have mentioned that this could be a product on its own. If I were to scale it up and turn it into its own startup I'd want to make sure not to saturate the tweetosphere with it or there's no benefit over the competition. I'd also probably license it out to early stage startups instead of bigger companies because no one wants to be spammed to by the next HP, Apple or Dell computer when you ask for suggestions from friends.
The problem is that Twitter's filtered Streaming API has quite a small limit on the number of words that you can provide to match tweets. So, it would be impossible to scale.
A better option might be retrieving tweets from the Firehose Streaming API, but one would need a massive backend to handle it. Additionally, access has to be requested to Twitter and it's very rarely granted. Alternatively, Datasift or Gnip can be better sources for the firehose, but I think they only provide a fraction (like 1/10) of the real amount of tweets that go through the actual firehose.
The Twitter Streaming API allows you to track up to 400 keywords. For most businesses, that seems like enough. Or if you need 1000, you simply need 3 accounts.
I'm already a fair way into building a similar paid tool (curated engagement via twitter+disqus), if its something you might want to try out ... email in my profile.
A question of ethics. I could have written it to be as flexible with its replies as possible but I didn't want it to look like spam. Even now the odd tweet gets passed but it occasionally makes for a laugh as the replies I'm giving make no sense based on the question.
To me, this is a big no-no. If you're answering someone who posted a question and it's clear that you're the business behind the product/answer that's okay with me. But if you're using a personal profile (real? fake?) to push traffic to your website, you're actively misleading people.
I think the question is, how spammy would you think if 1000 other people would do the same? How relevant would the answers be if every answer would have 100s of answers from bots that are advertising their service?
I agree with the sentiment that this is borderline unethical, but I think that it's probably okay as long as there's a human in the loop presumably spending as much time/effort verifying the usefulness of the tweet before sending it as it costs the end user to read the tweet and decide if it's worthwhile.
People often write about making web stuff here. And their email addresses are often easy to find. Does that mean that we're all asking for outsourcing firms to send us "manual, personal" form emails via a 99.8% automated process, even if a marketroid blesses each one?
Of course not. That's spam. Similarly, if I post a question on Twitter, I'm asking my followers. If some random jackass butts in because it's profitable for them, that's spam too.
There's a difference between holding up a 1x1 foot sign at a gathering of friends and planting a 1x1 foot sign on your lawn.
Are you expecting your tweets to be private or public?
If you publicly announce a problem, (and asking for public comments back.. ) then it is not spam if you get a relevant and appropriate message sent to you.
Scaping usernames here however is not comparable to the blog post's example
edit: misrepresentation as 'just another follower' is what i am against here.
It was what I originally was planning on doing but felt that a personal response would be better than a bot. During the hours of 3am - 9am I do set it up to be auto replies but there's not much traffic anyways. There are similar Twitter bots that do this and I've spoken to someone from Twitter about a grey of the ToS area I might be able to work in.
If this trick is still really working for you, please, do yourself a favour and shut up about it. Drawing attention to it is not going to end well for you, either twitter shuts you down or people just copy your idea and it loses effectiveness. The correct time to write about these things is when they stop working.
I agree, but the general point is that specific marketing tricks that work well should be hoarded, not shared, because sharing them destroys their value.
Because it'll lower your competitive advantage? If your main competitive advantage is knowing about some cheap Twitter trick--sorry, I mean "Growth Hack"--then you need to rethink your game plan.
If all he's doing is parsing search results, then nobody would be seeing his advertisements. You conveniently left out the part where he uses a bunch of different Twitter accounts designed to appear to be different people.
Great innovative lead-gen. I'm more curious about your database of destinations. Do you have formal partnerships with the sites you're linking-to to syndicate their listings?
We do. For most for most of the providers we have arrangements that give us commission but they're all providing us with an API or data dump of some sort.
I wonder if they are using paid interns or unpaid interns? Makes me wonder if the cost of paid interns was put towards regular advertising whether it would generate a better return.
My account had posted it from Medium earlier but I realized how SLOOW it was loading so I loaded up a Ghost Blog droplet for Digital Ocean and reposted it there. I needed a new account to post it on behalf of so a smashed my keyboard. Voial!
The reason you didn't address what phea said is that you promoted this story through abusive techniques. So abusive, in fact, that we've banned your main account too, as well as this site.
Edit: We're also banning the site that you've abusively promoted here in the past [1].
I'm not affiliated with it, but LeadSift does something similar to this and is actually pretty good at it for picking up new lead opportunities http://leadsift.com/.
This post is doing something a bit different and probably more useful, but it reminds me of the people that search Twitter for mentions of their competitors and then try to convince those people to use their product instead. I see this every once in awhile when I search for our product mentions and it makes me sad. I'm sad because it's a huge waste of time for them and they are actually damaging their brand (not ours).
Even though Twitter is public, there's still a sense that some random person isn't just going to start replying to your tweets.
Am wondering if there are an API based services which can be used for NLP? Something cheap because the number of messages to be processed will be extremely large.
The idea of directly engaging potential users actively seeking/asking for a particular product/service on Twitter is very interesting; but unfortunately the way you are implementing it right now is unethical and in violation of Twitter's Automation Rules and Best Practices [1]:
"Creating serial or bulk accounts with overlapping use, however, is prohibited."
I initially thought you were using your company's own Twitter account for messaging people; instead, by your own admission [2], you have resorted to using multiple 'realistic' accounts:
Well you can always argue what is right and what wrong. Very true, it is a violation of the Twitter terms of services but I still appreciate it for sharing with the HN community.
There has also been a post about Darkmarket (https://github.com/darkwallet/darkmarket) on HN. Again, I do not encourage trading illegal goods at all but as a proof of concept it is still impressive.
Perhaps, but some famous startups got traction by doing such "unethical" things. IIRC even Reddit seeded their site with fake accounts and conversations.
Also if the Twitter accounts are truly of the interns like the post says then it's not really that bad. In that case, it would really be an employee mentioning their employer's product.
89 comments
[ 2.9 ms ] story [ 70.7 ms ] threadRight now there is a big disconnect in the post. I am guessing its because you assume the reader understands even what ner/nlp is
But frankly, having seen the travel space up close, I find this system more interesting than your startup...
I have been trying to preview it, but the it get stuck after Enter keywords. https://www.dropbox.com/s/jr3kkq1x4p6wf3q/Screenshot%202014-...
Edit: I Read the post properly the replies are manual not automatic so I don't twitter will have any issue with it.
Per Twitters Rules for Identification of Spam[1]
-If your updates consist mainly of links, and not personal updates;
-If you send large numbers of unsolicited @replies or mentions in an aggressive attempt to bring attention to a service or link;
[1]https://support.twitter.com/articles/18311#
A better option might be retrieving tweets from the Firehose Streaming API, but one would need a massive backend to handle it. Additionally, access has to be requested to Twitter and it's very rarely granted. Alternatively, Datasift or Gnip can be better sources for the firehose, but I think they only provide a fraction (like 1/10) of the real amount of tweets that go through the actual firehose.
I'm interested in how that works. Is it a realistic-looking spam account you can hire? (Not judging - genuinely interested.)
https://twitter.com/DeasesRuthann/status/464387770178215936
https://twitter.com/DeasesRuthann/status/464082317913243648
https://twitter.com/DeasesRuthann/status/463863787963949056
She or he is supposedly in the UK and they "know a few places in Monroe"?
Some of their "tweets":
"About ready to knock doc the fuck out if if he keeps fuckin with these grades""
"Boost Male fertility http://bit.ly/aA4gur"
"I can never use anyone either I got it or ima wait on it frfr and that's that I really don't need nobody for shit"
"Miley Cyrus USED to be fucking stunning"
"Them was some good ass fights people who was there got the best view we just tried to include the outsiders and yall being ungrateful he'll"
Congrats on being a worthless spammer.
People often write about making web stuff here. And their email addresses are often easy to find. Does that mean that we're all asking for outsourcing firms to send us "manual, personal" form emails via a 99.8% automated process, even if a marketroid blesses each one?
Of course not. That's spam. Similarly, if I post a question on Twitter, I'm asking my followers. If some random jackass butts in because it's profitable for them, that's spam too.
Are you expecting your tweets to be private or public?
If you publicly announce a problem, (and asking for public comments back.. ) then it is not spam if you get a relevant and appropriate message sent to you.
Scaping usernames here however is not comparable to the blog post's example
edit: misrepresentation as 'just another follower' is what i am against here.
If you were to use the same procedure but send the replies off using a bot and the API, then I believe that it would be against the Twitter ToS.
First the product, 'let's aggregate the aggregators!!!'
The monetization: those affiliate marketing codes everyone else uses!!!
Then there is the user acquisition strategy. Getting some intern to post stuff to Twitter for you.
The spin off product: a Twitter spamming machine!
Then there is 'sharing economy'. Is this really all it amounts to?
It'll never last, they're going to disrupt us any second now.
... Why?
Because it'll lower your competitive advantage? If your main competitive advantage is knowing about some cheap Twitter trick--sorry, I mean "Growth Hack"--then you need to rethink your game plan.
Per company security policy you have been denied access to the website: http://trevorstarick.com/twitter-an-untapped-well-of-potenti...
Reason: Not allowed to browse this Botnet site.
Edit: We're also banning the site that you've abusively promoted here in the past [1].
1. https://news.ycombinator.com/item?id=7637010
Even though Twitter is public, there's still a sense that some random person isn't just going to start replying to your tweets.
Before we launched were were getting about 50 signups/day from Twitter at our landing page - around 150 visits.
https://www.mashape.com/jetlore-dev/semantic-text-processing...
Demo:
http://dev.jetlore.com/tech-demo/text-processing-api/entity-...
Disclosure: I used to work for Jetlore.
However, there are niche markets where this could be useful. This is especially true when the user has a limited means to pay for ads.
- How many of those users convert to cash in the bank and recurring customers?
- Has this actually resulted in a better ROI than other means?
- Does this sort of semi-solicited tweet advertising gain or lose goodwill?
From a tech standpoint, this is cool, from a business standpoint, it's not proven yet from this writeup. IMO.
I know a few people who worked on a startup doing exactly this some 2-3 years ago.
Problem was it would not really scale plus Twitter really limited the free pipe.
There were a lot of hotels interested, but not enough to pay.
"Creating serial or bulk accounts with overlapping use, however, is prohibited."
I initially thought you were using your company's own Twitter account for messaging people; instead, by your own admission [2], you have resorted to using multiple 'realistic' accounts:
https://twitter.com/search?q=outpost.travel
This kind of growth hacking is misleading and unethical, and I think it has no place in Hacker News.
[1] https://support.twitter.com/groups/56-policies-violations/to...
[2] https://news.ycombinator.com/item?id=7720634
There has also been a post about Darkmarket (https://github.com/darkwallet/darkmarket) on HN. Again, I do not encourage trading illegal goods at all but as a proof of concept it is still impressive.
Also if the Twitter accounts are truly of the interns like the post says then it's not really that bad. In that case, it would really be an employee mentioning their employer's product.