Hacker News new | past | comments | ask | show | jobs | submit login
Using chatbots against voicespam: analyzing Lenny’s effectiveness (acolyer.org)
97 points by mnem on Sept 10, 2017 | hide | past | favorite | 31 comments

Lenny is the most effective time waster for telemarketers I've ever heard. He's truly awesome.

I have seen the reverse problem, where the spammer who called is trying to get off the phone. Many years ago, I had a job as a manager in an outbound telemarketing center. We had a policy that we could not hang up once the call started -- the person we called had to hang up. Supervisors took escalated calls where someone whose dinner we had interrupted was upset that we called. Very rarely, those calls would then be escalated to me.

One time a lawyer in Philadelphia (not even kidding) refused to hang up, and instead continually berated the rep on the phone with cursing. He refused to hang up. The supervisor couldn't calm him down, and had repeated all the stock responses we had apologizing for upsetting them and promising not to call them back. He just kept cursing and said he was going to stay on the phone all night since we couldn't hang up. After about ten minutes of this, the call was escalated to me.

I went through the same items with him when he paused to take a breath. He chuckled and said he was going to keep me on the phone all night. I told him that was fine, since I was paid hourly and had gone into overtime five minutes earlier. (It was five after the hour.) I said, I'm now making $22.50 an hour listening to you. I'm happy to stay on all night. How much are you getting paid? That made him hang up.

These days, though, most call centers have a policy where the rep on the phone can hang up if the customer continues to curse after being asked to stop. Requiring the rep to stay on the phone beyond that can get an employer in trouble for creating a hostile work environment.

I keep them on the phone long enough to say a name, which company they are from and what they are trying to sell me.

Then I tell they are breaking the law as I am registered with The Telephone Preference Service [1] and they are on breach of the Privacy and Electronic (EC Directive) Regulations 2003.

I have had one successful prosecution so far.

[1] http://www.tpsonline.org.uk/

heh, I might have just left the phone going, that 22.50 an hour was coming from somehwere...

Yes, telling me the call was costing them at least 22.50 an hour would incentivise me to stay on the call for as long as possible.

For some reason, that made all his rage evaporate. And I really wanted to go home. That kind of work is emotionally exhausting.

That kind of work is emotionally exhausting.

As well it should be.

I completely agree. Much as ex-smokers despise smoking more than those who have never smoked, I despise marketing and advertising.

I don't want to know about your new product or what it can do for me. I've been living quite well without it so far, thank you, and now please FO.

I would pay $0.01 a minute to forward calls to something like this. (Up to $1/call or something)

I'm invisioning a marketplace of chatbots trying to pass the turing test (ideally armed with cancelled credit card numbers, fake postal addresses and bogus identities -- and willing to do absolutely anything to waste the callers time).

The beauty of it is that normal spam calls have assymetric economics in favor of the spammer.

A team of chatbots turns the economics around.

$2/month or $6/year.


You can pick a specific chatbot: http://www.jollyrogertelco.com/pick-a-robot

I wish there was a service like this in Australia.

Aftee some years, the spammers will all be robots too, leading to an amazing world of robots arguing with robots over the phone, both pretending to be human.

Almost all the cold calls I get now are bots where you have to press some button or say some keyword to get to a real person.

Sometimes I like to request a human just to waste their time. Organizations shouldn't be able to harass people for free.

I see this akin to adblockers that generates fake clicks. The sentiment here is if we can waste their time (and money) at scale, it's all for the better.

But with fake clicks, I often read on HN that it is the wrong way to respond. Can someone explain the difference?

I don't have an opinion on this technique with ads vs. calls, but ads are transactional. You get some piece of content or an app in exchange for your attention, usually initiated by the user, and with low friction to close the tab or uninstall the app. With calls, the moment it rings you are interrupted, the other party reached out to you, and you get absolutely nothing in return for your wasted time.

Ethics seems to be the big reason.

Here you are getting hit by an unwanted ad over hardware you pay for.

Adbockers are an attack on hardware and IP you are using that someone else owns and pays for.

CNN content and servers are 'free' on the condition you view their ads. You are going to them, they aren't pushing on to you.

The guy whos runs that bot is super weird about sharing his recordings/code. I'd love to implement this on my PBX but he won't share to "prevent abuse" (unclear what abuse he's afraid of). Idk, it sketched me out.

If you want to implement Lenny yourself with FreePBX and copies of the Lenny audio files: https://www.crosstalksolutions.com/howto-pwn-telemarketers-w...

While these conversations are undeniably hilarious to listen to, it's a little concerning to me that by the end of the conversation the only explanation you could really come to here is that he has dementia. If this countermeasure becomes widespread, I'd be concerned about unintended consequences for actual humans with age-related cognitive decline.

For example, imagine an older person who easily loses track of the conversation picking up the phone to talk to a telemarketer. The telemarketer mistakes this person for a bot, and yells at them in frustration.

This might be mitigated by adding some actual AI, so that the bot can convincingly imitate a person without mental health problems.

The telemarketer assuming they are talking to a bot is possibly a better outcome for the older person than the telemarketer taking advantage of their mental state.

Well, it quite helped me. At least I don't need to talk to them. However, some of them still insist on calling me. I think legal approach is the best way for us to make them stop, like what I read at http://www.whycall.me/news/consumer-wins-massive-229500-robo... about a well-known company which has been sued by a consumer because of telemarketing calls/robocalls.

Lenny's author wrote a lovely description of how he designed Lenny (skip to "If you're still reading, you might be interested in Lenny's background"; a few paragraphs after explaining why he doesn't want to open source the voice recordings):


I have grown very fond of Jolly Rodger Telcom (http://www.jollyrogertelco.com/). I will have to check Lenny out.

Listing to the demos they aren't really boots but rather just pre-recorded sound bytes. Lenny actually changes it's response depending on what the caller does. Pauses, keywords, etc.

> Lenny actually changes it's response depending on what the caller does. Pauses, keywords, etc.

It's actually a sequence of recordings played in the same order every time in a loop from the 5th. Telemarketers usually realize it a few recordings into the second loop.

Now imagine combining machine learning with conversation analysis to create the ultimate Lenny.

You should listen to some on youtube, they are not simple loops


one of them keeps trying for over 50 minutes

They are loops. I just listened to the 50 minute one, and 35 minutes in he's saying the same phrases he's already said near the beginning. "3rd eldest daughter..."

The article says the opposite.

"Lenny simply plays a set of audio recordings one after another to interact with the caller. The same set of prompts is always used in the same order."

RoboKiller.com has an app that intercepts robocalls and uses bots to waste the callers time. It feels good to get payback on those bastards.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact