> Honest question, why can Elon, and only Elon solve the bot problem?

I don't think _only_ Elon can solve this problem, but one way of looking at the problem is that Twitter appears to be hesitant to 'solve' the bot problem because their most important metric that drives stock price is the number of engaged users, which includes bots. Taking twitter private (is this actually what Elon is doing? I haven't been following very closely) potentially removes the need to focus on this metric, freeing them to 'solve' the bot problem and not worry about the effect on the stock price when the # of engaged users drops by 40%.

I worked for a social media management app company (Sprinklr/Hootsuite/Sprout-esque) and we had the exact same problem. My ballpark estimate is that at least 90% of the messages that went through our system were extremely obvious spam or scams. However, there was a huge focus on MAU growth, so there was no way in hell we were going to attempt to combat this.

For legacy implementation reasons, the backend service that published messages to Twitter would lose 1-2 queued messages every time the service was redeployed. Bleakly, the engineering team decided to not consider this a major issue, despite it being a core competency of the platform. It was simply extremely unlikely anything of value was being lost.

Did you try to introduce a new measure "MAU excluding bots"? Then over time the company could refocus around that and better policies be made

That seems like a good idea, but I suspect introducing a "MAU excluding bots" metric would look to investors like a sudden drop in MAU by 90%.

It would seem reasonable for large investors, -those who have influence on a company, to require MAU excluding bots figures as part of their due diligence. On the other hand: FOMO, guess.

The only economic actor directly hurt by bots is the advertiser, right? To some degree, it's against investors' incentives to attempt to strictly exclude bots.

Obviously in the longterm it hurts value, but "doing things that don't scale" creates a lot of degrees of freedom (user growth, cashflow, funding) that you can use to pay off your "tech debt".

Thanks to this thread I now see why the bot problem is sad difficult to solve. It sticks deep. Damn.

I wasn't nearly senior enough to have any influence in something like that, but in the internal culture, MAU was seen as such a strong proxy for success that suggesting the metric was flawed would have been sacrilege. The company had big parties and put out press releases every time a multiple of 1 million MAU was hit. To point out that most of those users were bots abusing our service's free tier would be noticing the emperor had no clothes on.

Then more bots come in undetected and you have to rebase again when you find them. It's a continuous process. You need very understanding investors to pull this off. At the same time there's only so long you can sweep inauthentic engagement under the rug until someone calls you out for it.

I don't understand why investors don't ask for numbers that exclude bots. Or maybe they do, and we just don't hear about it? This isn't some unknown problem, right? Bots have been around forever and they are a problem in many platforms, not just Twitter-like platforms.

The core problem them was a lack of integrity on the part of the executives. Which I am not saying they were especially unethical. Lack of integrity is very common in executives. But still, that is the root cause there.

Is it possible that Facebook’s apps could also be using bots to boost their DAUs etc?

This nailed it. Public companies have to "grow" and if growth is bot-driven then combating it is not goin to happen. By taking the company private Elon could (in theory) throw out MAU metrics, engagement metrics, etc.

He could "Make Twitter Great Again" by killing the bots, and rage-engagement, and making Twitter interesting and fun again. If that raises Twitters value by 10% he gets a $4.5B return.

Everything I've read says he's taking it private. Yes.

Doesn't advertising revenue require the same metric to be high? Assuming he wants to make money or just be self-sufficient, not operate at too much of a loss.

If you can improve the general signal/noise ratio I think that'd be big for advertisers. I think it's one of snap's strengths actually.

If Elon makes it subscription based in addition to being private, I’m confident that the bot problem will be solved.

Publicly traded companies are too often affected by short sighted investors.

Twitter is not "trying to solve this problem", they are embracing it and benefiting from it. They are strongly incentivised to keep the bots because it gives them a massive boost in their number of active users and the valuation of their company.

I don't think this would ever change unless they go private with someone that has no intention of selling it. I have no idea how successful Elon would be at removing them, but at least he intend to try to do it.

I don't think that the biggest benefits that Musk is bringing are technical ones, although he can surely afford to mobilize a lot of technical power, but policy ideas. Twitter will be run differently, become less toxic, and hopefully in addition to that game-changing attitude the bot problem can be addressed as well. I don't see how it can get worse, but there's a lot of potential for it to get better. Who knows? I might actually start using it.

> Twitter will be run differently, become less toxic, and hopefully in addition to that game-changing attitude the bot problem can be addressed as well. I don't see how it can get worse, but there's a lot of potential for it to get better.

While I'm an optimist in real life, unfortunately I don't have much reason to believe anything will improve, particularly due to Musk. He is hardly the most reasonable or grounded person. (Mind you, as a twitter user I do hope it improves.)

Overnight it’ll become less toxic because he’ll “run it differently”? I’m skeptical.

> surely afford to mobilize a lot of technical power

How do you reckon he’ll do that?

I think one of his strongest skills is recruiting young engineers who are ready to work harder for him than they should for less pay than they could get elsewhere.

And not just twitter!

Surely the biggest outgoing for marketing departments is paying for 'active users'.

Are you suggesting fraud? If not what? (I also suspect many MAU numbers are fraudulent)

Edit: spelling

Elon isn't paying for this out of his own pocket. He's not transferring it to some perpetual trust a la the Guardian and the Scott Trust. He's taking it private but there will still be pressure from everyone he's raised money from to make sure they make good.

Mostly, I think this is going to end up being one of the most expensive acts of individual hubris in history. It's strange from someone whose other notable projects have such clear goals. "Moar free speech lmao" doesn't sound measureable, let alone achievable while also making bank.

The money he's secured is mostly loans, not venture capital, so that sort of pressure to grow or pad numbers should indeed be diminished.

Bloomberg mentioned two loans to cover the shortfall in his equity swap (one external and one drawn on his Tesla stock), with annual interest payments in the ballpark of $1B each. So Twitter's take home pay just lost $2B of the $4B it nets annually. Losing further market share by re-engineering retweet algorithms to be kinder/gentler looks like a better bet to bury Twitter than raise it.

But mature media giants are supposed to lose money, right?

"one of the most expensive acts of individual hubris in history" is what people have been saying about everything Elon does for a while now.

> "Moar free speech lmao" doesn't sound measureable

Did you just make up a ridiculous quotation and then attribute the sound of it to Musk ?

>why can Elon, and only Elon solve the bot problem

To me it seems that he's the only one who wants to.

Bingo. Google and everyone else makes money on volume, not quality of content. They don’t care if the majority of the volume of content is garbage as long as they get their clicks/impressions on ads.

Reposting this because unsure why it was flagged, as it is perfectly valid and based in objective reality:

They (twitter) remove a lot of volume because it doesn't fit with their political narrative, even though this is financially detrimental to them.

What's "a lot" of volume in the context of this claim? Do you have a source for the proportion of would-be active tweeters they block, or proportion of tweets they take down, or anything like that?

Donald trump had a large amount of followers and did not threaten violence beyond his role as us commander in chief, much like his predecessors and successors have.

In this case you don't even need a source. You just need to have your head out of the sand, but outside of that, you can look at all the users who have left the platform or became entirely inactive after Donald Trump was banned, then look at the surge of users the alternative Twitter platforms gained, and pretty much make a safe assumption they lost at least that. You can also take a look at Twitter's stock price in Jan 2021, where it was close to 77 dollars, and rising at the time, and compare it to the pre-Elon hype price of 39 dollars, with Goldman rating it as being closer to 30 dollars in true value.

> you can look at all the users who have left the platform or became entirely inactive after Donald Trump was banned, then look at the surge of users the alternative Twitter platforms gained

Do you have those statistics on hand? I don't, so I was asking.

I know the stock price is substantially lower than its peak 1+ yrs ago -- it's easy to find a graph of that -- but it's not obvious to me that this is the result of Twitter's politics rather than other factors.

Trump was prompinent on twitter and violated their TOS for years. He was banned on Jan. 8th. If Twitter were protecting some political narrative, it wasn't reflected in their ban.

This argument holds zero water. You have Iranian officials openly calling for revolution, Chinese foreign officials spreading literal propaganda, and Saudis calling for actual executions, which I would think also consistently violate their TOS, and yet have seen no similar measures taken against their accounts.

That only demonstrates lax enforcement, not that Twitters mission is pushing politics in lieu of money and active users. Trumps politics haven't changed since he started his right-wing schtick.

It does seem odd that the example of enforcement is a very interpretative reading of a center right view, and the examples of non enforcement are every other direct call to violence.

I find the violating TOS argument for why Trump was banned flawed because the TOS allows for very vague interpretation. For example the tweets that supposedly got him banned said something to the line of “fight for democracy” and that was interpreted as encouraging actual fighting and the illegal activities on January 6th. It also ignores the fact that he said along the lines of “we need to go and march peacefully, don’t cause problems, that’s exactly what they want” and as everything was happening was tweeting that everyone needed to respect police, be peaceful and go home. Good luck even finding the tweets from January 6th because only a couple outlets even reported what they were, only that they violated TOS which is also very suspicious. Edit: typos/grammar.

Does he? I haven't followed everything he said about this, but I thought he mostly wanted to undo bans. I haven't heard him about the bot problem.

Meanwhile everybody else does seem to see bots as a problem. So even if Elon wants to, I doubt he's the only one.

He wants to charge for a checkmark. This both increases the cost of bots and is a form of authentication. It’s much easier to compare payment methods between accounts.

>He wants to charge for a checkmark. This both increases the cost of bots and is a form of authentication. It’s much easier to compare payment methods between accounts.

What makes you think users want any of this?

Frankly, most of the suggestions I've seen coming out of the SV "thought leaders" about what to do with Twitter are terrible. The bot problem isn't easily solved, and adding a ton of friction and/or reducing anonymity isn't going to be some huge boon to the platform.

I've often thought that if they 1) allow anyone to verify their account and 2) let verified accounts choose to only (or mainly) interact with other verified accounts, it can make it more obvious which accounts are anonymous or bots. I think it should work unless/until the anonymous and bot accounts figure out how to break the KYC process.

> 2) let verified accounts choose to only (or mainly) interact with other verified accounts

That is already a thing if you are verified and one of the reasons why "blue checkmark for $2" (Elon's plan) won't work and will make many verified people unhappy. They don't want even more peasants and "cloud chasers" in their verified timeline.

Hence different color. I also don’t agree. I think most blue checkmarks equally hate the spam and anonymous hate. In a public square, we can tell you’re a robot or if you’re not and say vile things, you’re unlikely to stay anonymous.

Yes, I'm all for adding layers of social friction, or frankly, reducing the emotional distance. I remember reading a book, I think called On Killing, talking about the psychology of war and violence. In it, he mentioned how it gets easier to kill from a larger emotional distance. Or in other words, it can be harder and harder to harm someone the closer we feel to them. Being able to put on a digital mask and act as a persona can lead to this (it can also lead to opening up more, too), but I at least want to know if someone is wearing a mask or not.

I think the challenge is "verified" = celebrity, or verified fame, and really I want it to equal verified humanity. Maybe that means the verified humanity gets a different name, who knows. There could be two tiers, allowing the verified famous people to interact with each other and others to just know who went through the KYC process.

Yes. But the opacity on how to get verified is ridiculous. And Twitter has used revoking a blue check mark as a punishment. As if the person that was once verified is no longer verified. It’s ridiculous.

Yeah, it has confused me that it's easier to verify one's humanity on dating apps than it is on Twitter.

Right. It’s like removing spam.

Frankly, I think Reddit could do the same and give more weight to upvotes from those that prove they’re real.

It could help with disinformation problems and with finding signal.

I remember reading an article where the CEO of Myspace said one of the main reasons Facebook won was because of its real name policy.

<edit>I found the quote [0], which I had quoted in my essay "Why Twitter Should Verify More Users: Towards a More Human Web" [1]:

> Facebook’s killer feature was that it replicated the real world by forcing people to use their real names, whereas MySpace users used pseudonymous handles

[0]: https://www.yahoo.com/news/former-myspace-ceo-explains-why-1...

[1]: https://medium.com/hackmentalhealth/why-twitter-should-verif...

</edit>

I know one of the things I loved about Facebook was interacting with university-verified humans. Then it got flooded with more and more spam accounts.

I hope either these platforms or the next generation of platforms realize there is a strong opportunity for real-name policy platforms. It won't be the whole market, but I'm personally really tired of anonymous accounts. I mostly come to HN because there aren't many real-name platforms that talk about these things. Maybe that's because lots of people don't want it. I guess I just don't buy that argument. I think there's a niche for sure.

Interesting. Thanks!

Go follow all the big crypto/NFT accounts on a fresh twitter account. You'll be able to see all of the easily solved bot abuse.

Elon said this. I don’t know what anyone else is saying.

He must think it will work?

1. Retailers have had issues curbing bots hoovering up video cards, PlayStations, etc.

2. Most of the Twitter bots people are worried about when they talk about bots are state-funded.

If companies can't prevent a self-funded scalper, what makes you think Twitter will be able to prevent Russia? There are a lot of ways to obfuscate payments so they don't look to be coming from the same source. Even more ways when you control the banks.

Retailers didn't really care. A sale is a sale, and scalpers are less likely to return or ask for a refund.

They care a little, because establishing relationships with (and gathering data on) real customers is more valuable than having single entities hoovering up all the supply. Further, real people often buy more than one item. I'm sure you've noticed retailers putting limits on specific items before.

> Further, real people often buy more than one item.

But... scalpers do this, especially so. As far as a retailer is concerned, a scalper is every bit as 'real' as any other customer.

1. Retailers have had issues curbing bots hoovering up video cards, PlayStations, etc.

I work in this field and to be honest, the main reason retailers have issues curbing bots is that they don't really spend much resource on it. It's a game of cat and mouse, there's constant work that needs to be done but there's a lot of low hanging fruits that the vast majority of retailers don't do to block bots. And, no, just setting up perimeterx and calling it a day is not enough.

I agree with dannyw. Also, Twitter influencing requires scale. You can fake some but patterns will emerge that make it harder and harder.

Now, if it succeeds, that does create an incentive for corruption within Twitter. The enemies of democracy have huge incentives to keep up these divisive public influence campaigns that have worked this past decade.

I find this to be a great idea personally. I will totally give Twitter a few bucks a month for less bot BS and a blue checkmark.

...and your government ID?

You give the grocery store your government ID when you buy oreos?

Well there are the cameras in the parking lot logging my license plate, facial recognition tracking my course through the store on the way to get the cookies and to pay for them, probably at a self-checkout terminal with a camera right in your face. Then you pay for it with your Visa and that Oreo (automatically capitalized by my iPhone) purchase data is sold by Visa. Then that data gets consolidated with the data secured from Verizon and their Custom Experience badged reading of your text messages.

Then THAT gets sold to the Depts of Defense, Homeland Security, and whoever else wants.

I feel it’s fair to say we do.

huh?

Yep, as long as they restore Trump and BabylonBee i might give Twitter another try

How does charging for a checkmark prevent bots? Forgive me if I'm wrong, I'm not a twitter user, but typically, aren't bots not verified?

The idea is that you can filter out the unpaid.

How many accounts are needed to meaningfully influence a Twitter story? How much can you charge for a checkmark that people will pay for?

Because if a checkmark costs $1/month for example, then spending a couple grand to be able to turn any story you want into "trending" is chump change for corporations, governments, and any particularly motivated silicon-valley salary'd employee.

There are already services which essentially sell you retweets - all this is is an adjustment to their business model to sell "verified retweets" and "aged verified retweets" (in case you want accounts with a posting history to do it).

It's a cost model adjustment, that might not even cost that much - after all if "verified" accounts receive less scrutiny from anti-bot algorithms, then they'd be cheaper to run then the developers/workers who need to keep your sockpuppet farm ahead of the game.

What if it cost $2 a month plus a scanned copy of a government-issued ID?

It’s possible to generate realistic looking IDs. It’s not that hard, any person reading this comment can do it.

Get a face from a thispersondoesnotexist.com like generator. Get the format of the ID in a template, fill in random but realistic data. Like for example, 1000 names for each race-gender combo. Realistic zip codes and addresses from another source.

It’s about a weeks work to get it working to production quality, but it can be easily done.

For bonus credit, see if you can generate an image of the same person holding up this ID. Harder, but not impossible.

What’s frustrating is, those of us who work or have worked in Integrity have already discussed all of this for years. We know the easy solutions, the low hanging fruit. It’s not that helpful when a bunch of people say “why don’t you just charge $2 a month”

Governments can generate IDs for thier sock-puppets for online influence campaigns.

Non-government actors can get in on the action by procuring IDs in other ways ("Make money by scanning your late maw-maws ID"[1]), or large scale phishing campaigns (fake jobs or college admissions abroad with 100% applicant success rate, the only catch is scanned IDs and qualifications are required).

Electronic forgeries are likely the best way to do it at scale; just the other day, therr was a top story ok faking as if your document was scanned. Do the fine people handling verifications at Twitter know what a legit Madagascan ID looks like?

1. Or "Get $10,000 funeral assistance from this new Biden admin relief program"

Now you are committing extremely serious crimes instead of just spamming though, few would risk that. So this might not solve all spam, but it would probably solve 99.9% of it.

> Do the people at Twitter know what a legit Madagascan ID looks like?

Yes, there are guides for checking ID cards, how do you think that they check ID cards in real life?

> Now you are committing extremely serious crimes instead of just spamming though, few would risk that.

Oh no, I've broken the ID forgery laws of Burkina Faso and am in big trouble now. Isn't it a rule of low-level scamming to always target jurisdictions away from your own? State-actors don't care about commiting crimes against themselves, only the mid-level, "semi-pro" threat actors care about avoiding major crimes.

I still don't underatand why paying would help. Wouldn't simply filtering out the unverified work the same way?

Paying verifies.

Obviously, but verification already exists. For your answer to make sense, one needs to assume that there's a non-trivial problem of verified bot accounts.

Is that assumption correct? It seems incredibly wrong to me, but I may be misinformed.

Verification only exists for a small group not the general public.

Ah OK, so you're saying that this will enable an expansion of verification, which will then serve as a solution to the bot issue.

They aren't. There are sometimes hacked verified accounts being abused, and I guess charging for the checkmark would result in less verified accounts and thus less hacked accounts... but bit of a reach :)

Those are also not really the problem.

Send them a 2fa key token with their fee and call it a day

Once upon a time, I was on a dev team which had to share an RSA 2FA token for accessing a client's VPN. Rather than pass it around to whoever needed it, we just pointed a webcam at it.

Damn, now it'll be even more of a good idea to immediately block every bluecheck.

Who is going to pay? There's at most a few tens of thousands of people who care about their identity being verified on Twitter (because they're journalists, company leaders, politicians). How much is Twitter worth to those people? Probably not $10,000 per year, so we're talking about a very limited additional revenue source.

For a lot of other people on Twitter, the pseudo-anonymity is a feature not a bug.

WhatsApp charged a few dollars a year and did fine.

I’d pay if it allowed me to filter out garbage.

It’s like paying for the removal of commercials.

I don’t believe Musk intends to require people to be publicly identified.

I don’t believe I follow anyone anonymous on Twitter.

Twitter has over 7000 employees, for WhatsApp's business model to work they'd need to cut this by 10x or so and there is no way they justify the 45 billion valuation if they do that.

Huh? I’m not following. Why would this hurt revenue?

Authenticated real people are more valuable to advertisers.

Did WhatsApp have Ads? I thought all their income was from subscriptions?

If it’s only for verification they, yeah, they might be fine as long as the verification fee is optional.

I do not believe they are literally applying the WhatsApp model. The point was that small amounts of money are tolerable in a network effect.

Remember for a moment that we live in a world where people drop $20 for a virtual Sword of Wounding +4. Often more than monthly in a single game.

A few bucks a month is nothing these days for dedicated netizens who want their social drug of choice.

Most large companies have trouble breaking out of a local maximum. The incentives align to just keep climbing the gradient and never go down.

Elon has the cash + clout to solve the problem. Is he unique in the ability to solve it? No. But he may be in a unique position to solve it.

The prevailing opinion on HN seems to be that it can't be solved. Of course, less than a week ago the prevailing opinion on HN was that Elon wasn't serious about buying Twitter, and didn't have the money, and it was all fake and part of a pump-and-dump

Just see the top comments here:

https://news.ycombinator.com/item?id=31109355

TikTok does a pretty good job with its ridiculous security ecosystem hyperfocused on prohibiting any third-party clients.

APIs need to use actual headless browsers in order to generate the appropriate signature.[1]

Maybe s̶o̶m̶e̶ ̶o̶f̶ ̶t̶h̶a̶t̶ ̶4̶3̶B̶ ̶c̶a̶n̶ ̶g̶o̶ ̶t̶o̶w̶a̶r̶d̶s̶ ̶a̶ ̶s̶i̶m̶i̶l̶a̶r̶ ̶s̶y̶s̶t̶e̶m̶ the rise in valuation, and Elon's involvement, can bring in some more money for the company itself for these kinds of endeavors.

https://github.com/davidteather/TikTok-Api

This makes automated posting difficult, but not sock-puppeting. When people say "bots", they probably don't mean automated posting, they mean creating lots of accounts to swarm other users with. You don't need much automation to do this; a device farm of old phones and some ability to remote-control each one would be enough to maintain hundreds to thousands of sock-puppets.

Interesting, I wonder if VMs would also work for this purpose?

$1 per hour manual labour would be cheaper more effective than updating scripts against a determined adversary.

Android in a VM.

The 43B goes to Twitter's shareholders in exchange for their shares (and financiers who arrange the deal), not Twitter itself.

On the contrary, Twitter is going to be saddled with some debt that will be used to finance the takeover. But perhaps this doesn't matter to Musk.

Very true, that's my bad.

Hopefully the rise in valuation, and Elon's involvement, can bring in some more money for the company itself for these kinds of endeavors.

Notably yt-dlp is able to extract from TikTok well, without using a headless browser.

https://github.com/yt-dlp/yt-dlp/blob/master/yt_dlp/extracto...

Someone might’ve said, how is Elon going to make space flight far more affordable?

I have no idea, but he’s a better bet than most other folks I’d say.

Will he succeed? I’d bet not, but I can’t think of someone better to try.

If Twitter wanted to solve the problem they would have. Simple as that. Elon is the one who is putting his money where his mouth is to get the company to tackle the issues plaguing the company.

The only software devs that in a reasonable position to solve the problem work at large social media companies, and those companies are financially incentivized to not solve the problem. I don't think this is a software capability issue.

The problem is not a software issue others can't solve but Elon's team. It's a cultural issue that the current Twitter team accepted in their comfort zone because it fits their bias (that does not defend freedom of speech).

Meta answer: the ultimate point of money is that it gives you the opportunity to make decisions without convincing everyone that you deserve to make decisions.

Taking Twitter private will give him more flexibility to implement structural changes to the business which may take time to see returns.

He has experience with it but he's not the only one. I'm sure Max Levchin would be very capable of putting together a team and fighting off the bots.

Combating fraud was a key technical challenge in building PayPal and bad actors on that network had more direct financial incentives to write them because PayPal was transmitting money rather than social updates.

What experience does he have with it? He's a capitalist, not a sociologist, programmer or even economist.

Musk was CEO of PayPal and definitely has a deep background in programming. He made his initial fortune through writing video games and then another software startup called Zip2 (a forerunner of Map Quest and Google Maps).

Levchin was PayPal's CTO and is also deeply technical.

I thought he made his initial fortune by being the son of the owner of an emerald mine in Zambia[0]

[0] https://web.archive.org/web/20140802011449/http://www.forbes...

I thought Musk was fired from both paypal and Zip2 for incompetence? Not something to brag about.

If you’re genuinely interested in the history, Jimmy Soni's new book, The Founders, is a great starting point.

Thank you

What's wrong with being a capitalist?

Politics aside it gives him no expertise in moderation queue management?

If skills worked like official titles and acquiring one meant losing another, this would be a serious problem.

Fortunately, people can and regularly do gain expertise in multiple things! Just as learning how to program doesn't destroy one's previously acquired drawing or writing skills, learning how to manage capital doesn't destroy someone's ability to combat fraud or detect malicious bots.

I don’t know if you’re talking about levchin or Elon but please do a google search before throwing around communist propaganda that is blatantly untrue. For both people.

Part of the answer has to be that companies don’t want to solve it because it would destroy metrics that buoy ad prices. Charge per tweet or per account and the problem would be gone overnight.

If you talk about social bots influencing opinions that's a problem that doesn't seem to exist, or at least gets vastly exaggerated. Here's an interesting paper from a German scholar who investigated this and wasn't able to reproduce any of the findings of other research in that area. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3814191

Because it's almost impossible to differentiate between a person posting a tweet and a bot doing the same. There is nothing humans do on Twitter than bots cannot automate.

If this were truly the case, then bots wouldn't be a problem!

More narrowly, I buy that this is true for some segment of the tweeting population. I know this is provocative, but removing those false positives from Twitter may be a good thing.

I have a Twitter account that I use very casually (1 tweet/month, maybe), and I'm 100% sure that my activity is not replicable by a bot.

I'd no longer call myself an expert, but I used to work in conversational AI and have kept up with the research. The latest language models are incredibly exciting, but they're still not at the level where they can simulate a non-braindead tweeter.

The above is more of an amusing thought experiment than an actually-feasible approach. Aggressive removal of botlike users would have to be opaque enough that it'd cut against the free speech objective that Musk has been promoting. But I will say that, at a smaller-scale, the best communities I've ever participated in have had incredibly strong content neutrality norms alongside robustly-enforced civility norms.

It also seems an odd thing for him to be solving since a lot of the marketing noise for his companies is bot driven. Perhaps he's more interested in "solving" the problem in specific ways.

https://www.latimes.com/business/story/2022-04-12/musk-is-of...

Elon has the ability (and motivation) to look beyond next quarter. Bots increase “monthly active users” metric, so short-term thinking management likes them.

Elon can afford to think long-term.

Twitters stock price has been flat since IPO... I think the board and shareholders have been ok with them thinking long-term seeing as how they have 0 returns.

The board barely owns any stock, and all collect a pretty healthy salary for basically doing nothing. I think I saw Elon tweet eliminating the board would save $3M a year and looking at the chart of board members and their payments it seems about right.

Judging by their past actions they were far more interested in using Twitter as a propaganda platform than worrying about it making money.

My amateur opinion is that it's not very difficult but it hurts subscriber count, clicks, engagement metrics etc. If he takes Twitter private there will be incentive to fix this because there won't be shareholders to please.

On the other hand, why are bots not considered a result of free speech? I'm not free to run hogwild on your platform with bots? That's censorship!

(This was originally a reply to https://news.ycombinator.com/item?id=31154051. We detached it to prune the thread, which was getting too top-heavy.)

Lol Elon has no interest in solving a bot problems. Elon wants to allow the alt right back in Twitter after they've been banned and failed to start a rival competition.

This is the official death of Twitter.

In the same way that Elon, and only Elon, can solve self driving.

Because he has money and so can easily make everyone listen to him, unlike most other programmers who are stuck behind business/product/finance/management guys.

I don't think that he and only he can solve the bot problem, but with question stated this way - Tesla may currently have the best machine learning hardware on the planet. With people reports + second wave of hired people for marking what is spam and what is not, this seems like a fantastic place to shine for ML - especially when you have all accounts data and full graph available.

I would not bet on it - as far as I understand it, fighting spam is like 95% of what engineers at Twitter do and openly available ML on GPUs seem plenty strong - but it is a possibility.

Tesla isn't Stark or Wayne Enterprises. You know that?

Yes. But they have Dojo.

the bot problem can be solved if there's political will. Elon's power isn't being the guy who solves the problem (though he is pretty good at tackling technical issues directly), Its creating the political will to solve problems other people find intractable and clearing a path and providing resources for smart people to solve it.

frankly I'm jelous and wish I could do that.

I don't really think they are trying to solve this problem as much as being reactive to complaints.

> Honest question, why can Elon, and only Elon solve the bot problem?

He can't and he is unlikely to want to.

I can think of couple of reasons:

1. as a public company, they are afraid to lose so many accounts - see Netflix as an example on what could happen to their stock

2. bots are the future, at some point I think everyone will have a bot to post for them

Genuine curiosity, what would someone have a bot post for them that they wouldn't post for themselves? I can think of edge cases, and genuinely good use cases, but what is the general purpose use you're thinking of that would see everyone use one?

Some wars for public opinion, are lost and won on social media.

Those who post quickest have a higher chance of not getting lost in the noise and that is where bots will probably be used first by individuals as well.

With that said, hopefully this not the future and new ways of moderating and presenting information will be invented, but the way things are now, it's a slug-fest and the fastest and loudest often wins.

I'm genuinely curious as to why you think the bot would be posting "faster" to someone's opinion or statement than a person who can read the statement and respond as fast as they can type..

I can see people setting up some form of pre-loaded opinion/responses based on what they want to be seen as..e.g. if $OPINION_I_DISLIKE print $YOU_BAD else print $YOURE_RIGHT! but that seems like it's pretty limited as to completely not work in practical human communication

Why would you want to have a bot post for you? Once we have general AI available, there will be much better applications than posting for you on social networks.

He isn't the only one who can solve the bot problem.

because its not a technical problem. its a political problem

No, they haven’t been trying to solve the problem. They want the bots on there for financial and political reasons. It is easy to solve the problem - require identification to use the site. I mentioned the advantages of social media doing this about 10 years ago and was downvoted on this very website because of “privacy and anonymity on the internet is great!”.

Just to note, there are many reasons to not want to "require identification to use the site" that aren't related to "[wanting] bots on there". Cost, friction, safety/protection, etc. - the list gets quite long with a trivial brainstorm.

Also note: that's not the same thing as saying "Twitter shouldn't require identification to register" (though personally I don't think that it should). It's just to say that it's WAY more complicated than "bots vs. ID".

Privacy and anonymity are great and it doesn't even solve the problem until your ID is associated with some form of public key in every country.

Why is this comment not signed with your full name?

How do you feel about inevitable leaks and your identity being stolen?

I do think that would work but I think twitter probably could detect bot actions if they wanted to in a more nuanced way. Swarms of new accounts making similar comments has to be detectable on their end.