Hacker News new | ask | show | jobs
Twitter Acquires Magic Pony (blog.twitter.com)
128 points by rogerfernandezg 3650 days ago
15 comments

Machine learning on your images and videos to "enhance our strength in live and video and opens up a whole lot of exciting creative possibilities for Twitter"!

As ironically stated here [1], "The best minds of my generation are thinking about how to make people click ads."

Though Twitter has a good record of open-sourcing internal infrastructure projects, I think it's it's safe not to expect any substantive amount of this work to see the light of day.

On the other hand, the quality of their search and recommendations when it comes to non-text content will significantly improve. They're really doubling down on Twitter being the "best place to see what's happening and why it matters, first", which is a welcome change from the confusing signalling in months prior.

[1] https://news.ycombinator.com/item?id=11930354

> "The best minds of my generation are thinking about how to make people click ads."

The best minds of the previous generation were thinking about how to kill Russians. I'd say it's an upgrade.

>The best minds of the previous generation were thinking about how to kill Russians. I'd say it's an upgrade.

At least the results of the previous generations research went (largely) unused, whereas the results of ad clicking research is targeted against and used on people on a daily basis, to their considerable detriment. An upgrade? If so, it seems kind of dubious at best.

Only on HN would someone compare starting a nuclear war to click advertising.
Well only one of those two things actually happened (thankfully).
Global nuclear war sounds immensely more fun than the present.
To quote Douglas Adams: "No, not really."
Surely you're joking.
As somebody who lives a mostly ad-free life, I totally agree. I think it's somewhat hard for people used to ads to appreciate how pervasive and manipulative they are. And it seems impossible to get people in the industry to see that maybe, just maybe manipulating people for a living is kinda sinister.

Even if it weren't, it's certainly wasteful. Advertising is mostly an arms race. Pepsi spends to challenge Coke. Coke spends even more to maintain hegemony. Neither is informing consumers, the perennial econ-theory justification for advertising. We all know those products exist. If we banned advertising tomorrow, consumers would be no worse off, and all those people could be doing something actually useful.

A shit sandwich is better than a catered shit buffet, but that doesn't make it good in the slightest.
> "The best minds of my generation are thinking about how to make people click ads."

There is a strong argument that as long as they continue to publish and produce open source (which they have been doing) then you could think of ads as just a source of funding and tons of data to continue to push the state of the art. It can also be a stepping stone to other ventures [1].

[1]: http://www.wsj.com/articles/where-would-we-be-without-intern...

Since the last few generations used weapons research to fund scientific advancements, I guess this is progress after all!
Don't mind the huge military-industrial complex that has helped derail education, health care, retirement-funding, ...
Remember that the US military Active Duty + civilian jobs makes it probably the largest employer in the US. Not everything they do is great, but they pay salaries and benefits.
A government employer for the sake of employment... Marx would be high-fiving himself in his grave.
Side benefits from the advertising-industrial complex?
> "The best minds of my generation are thinking about how to make people click ads."

That's just half the truth. The other half of the best are thinking how to skim money off financial transactions.

> They're really doubling down on Twitter being the "best place to see what's happening and why it matters, first"

It's a little ironic that they announce this in a blog post.

Wouldn't the Twitter equivalent be a tweet that points to the blog post?
Not really. This isn't the sort of news that most people need to know "first", or alternatively "live."
Or "ever".
This is the actual source of the quote: www.businessweek.com/magazine/content/11_17/b4225060960537.htm
> As ironically stated here, "The best minds of my generation are thinking about how to make people click ads."

Thanks. Perfectly succinct summary of what I hate so much about today's business world.

Dear all advertisers: how about you stop trying to figure out how to make people want your product/service more, and focus on making it discoverable and clearly understood when the time comes where they finally realize they do want such a product/service and go seeking it out from someone? And if they never do end up wanting it, MAYBE THAT'S FINE and maybe it's time you stop selling it?

As I was reading your comment, I was thinking - "Why not? Why shouldn't advertisers figure out how to advertise in ways that comes somewhat close to what you are suggesting, but make you want to buy before you know you want something?" ;-)...

Sometimes I think we are in a difficult dichotomy -- we like Twitter and we consume it voraciously and love that it is free, but we don't want them to use advertising as a source of revenue even though that might be the source that keeps it alive for us to consume. And, for free!

For sure, we're a greedy, selfish, and short-sighted species. But that doesn't make the advertisers right for doing what people are implicitly asking them to.
People say 'the advertisers', but don't most people in the world work for a company that uses ads?

It seems that there's a mentality of compartmentalization, where people believe there to be 'evil ad companies' that somehow do nothing but make ads and don't actually sell products, nor employ people in the creative industry.

> People say 'the advertisers', but don't most people in the world work for a company that uses ads?

I don't think so? 90% of businesses sell to businesses, not to consumers, and in that world it's more about networking than advertising. (You could argue that that's just another form of advertising, but at the very least it's easier to avoid).

Just a crazy idea, but what if instead of TV/radio commercials and billboards and internet popup ads, products advertised themselves in the places they are sold (e.g. the sign next to a restaurant, or the side of a shampoo bottle on a shelf in the store), to explain very briefly why they're better than the competition? Most of the time, people will come across it organically, and that's healthy and good. But if discoverability is needed too, they can advertise in appropriate places, like restaurant listings, etc.
People like to feel righteous and superior. Whether they deserve to feel this way is an entirely different story.
I don't know if advertisers are the villains here, or at least they're not the only ones. People want things that are expensive to run and maintain (journalism, social networking services, accurate internet search), but don't want to pay for them. AFAIK, advertising is the only way to square that circle.
Because when they do want it, there will be several competitors offering it and from experience I know they will most likely choose the name they recognize from my prior brand awareness campaigns, and do little other research.
In even further irony, in this particular case, they're one and the same.
> "The best minds of my generation are thinking about how to make people click ads."

What are people doing to change this? What do you think needs done?

Asking seriously because I'm actively trying to figure out the answers to these questions.

Make an 'ask HN' to keep this thread closer to its original topic. I'm sure many of us would like to share our perspectives on the matter.
Frankly, the majority of the absolute very best minds of my generation are now tenure track associate professors. They received offers from substantially all of the best research institutions in the country.

Much of the very very good minds are thinking of ways to make us click more on ads.

I looked into these guys a month ago, they have a great team of young researchers.

The main application I saw was applying neural networks to video compression, improving compression by 3x or more (I guess by regenerating the lost detail?). Not sure if Twitter has much use for that but it looks like they have patented a bunch of other stuff too, and I'm sure will be a great addition to twitter's ML team.

Congrats to the guys - and hope they continue creating interesting tech in old blighty after the vesting period is over.

Periscope could use it.
Do you have a link to the compression stuff?
Pied piper?
It's almost too good to be true, isn't it?
I know nothing about Magic Pony, but I'm starting to wonder how much of this craze is a form of "keeping up with the Joneses", except with tech giants and ML/AI startups instead of suburban families and their cars, etc.
Its not, its a fundamental shift in how you power your applications. Not acquiring and building this talent at a major firm would be the equivalent of sticking to steam power in the face of oil.
This is amazing to me, not because it's a great purchase but rather because of the absolute tone deafness that Twitter is demonstrating.

People are not using Twitter because of many reasons and bad image compression is probably not one of them. Unless Twitter is planning some massive pivot what they should have spent their M&A budget on companies that will help them improve their growth situation directly (in my opinion).

Twitter's massive pivot is actually well underway:

- They bought Crashlytics and released Fabric.io, a souped-up Twitter SDK + Crashlytics + their bundled ad platform MoPub. The idea is you'll develop your app with this SDK and get crash reporting, usage stats, ad serving, 'Sign in via Twitter', Twitter integration, and all of this other stuff [1] for free.

- They bought Digits, which is a 'log in with your phone number'-as-a-service and made it free to use. It's now integrated with Fabric.

- They just bought Magic Pony which will help them better classify non-text posts so that their currently-awful suggestions are better.

Fact is, 'Twitter', the service where people post tweets, is no longer their primary growth area. It's their most externally visible, sure, but they want their SDK inside every app, and mine data and serve ads, just like Google and Facebook. This might actually work, because their SDK has some very nice things you might want anyway, compared to the other two that only give you identity, integration, and ads.

[1] https://get.fabric.io/kits

[2] https://get.digits.com

It would be super cool if Twitter let me write my own algorithm to decide what I get to see.

Or even better, let me "publish" an algorithm and let other people use it.

I am pretty sure this whole article is just a euphemism for a new drug that twitter execs are super into right now.
Twitter should spend some of the money getting someone to actually improve their web ui. Click a link to an individual tweet and read it. Now you click out of it to read the person's timeline. Here comes the janky timeline refresh. I mean shouldn't someone spend some time so it doesn't look like they had to pull the user's timeline?

I still get errors sometimes where I see, "1 new tweet" and it's the top tweet in my timeline, except here it is again.

The bit about the Twitter Cortex team reads like an MBA approach to product. If I just rub together some PHDs and engineers then magic will happen.
This is very sad news.

The #1 reason why I moved from Facebook to Twitter was that the latter service used straightforward algoritms for deciding what to show me. When I opened Twitter, I could be almost sure that the timeline would only contain posts from people I followed, arranged in chronological order.

Of course, there were some exceptions. Firstly, they always had ads. I see how many people here get angry about it, but I think ads on such site is perfectly normal. After all, they have to make money somehow, right?

Secondly, there were times when I saw "While you were away" box. This was the thing I found annoying from the beginning. How did Twitter infer that I was interested in these five posts more than in those five posts? It got worse over time. In fact, now this box shows up after a few hours of my absence, as if I am expected to check Twitter 10 times a day.

And it appears that they are planning to take these features to a new level. I really hope they don't do everything they talk about, but if they do, maybe I will have to seek a new social media service.

I can't laugh hard enough at this. I agree I (sorta) hate the Facebook filtering algorithm, but I find Twitter even more impenetrable. It's just impossible to figure out what's going on. Here's my timeline, just opened it up for this comment:

* Account I follow, 15s ago.

* PROMOTED post from 5 days ago. (here I've already run out of browser real estate on a 27" display because the tweets take so much damn room. Scroll down a page)

* While you were away: post from 3 days ago

* While you were away: post from 4h ago

* While you were away: post from 1 day ago

* While you were away: post from 3 days ago

* While you were away: post from 6h ago

* While you were away: retweet from 2 days ago

* While you were away: post from 1 day ago

* Actual twitter post from user I follow, 2h ago.

* PROMOTED from 18 days ago

For all intents and purposes, the Twitter feed is less chronological, less useful, and less content-dense than Facebook, and that's saying something considering that Twitter is supposedly a 140char service.

I had to scroll through 3 pages of crap before I got to the actual chronological twitter you speak of.

Also, they have this horrible habit of making tweets with a white background, then a tiny grey divider between tweets so it can be hard to tell at a glance when you've moved from one section of content to another -- much like when GMail went from colored message threading to all-monochrome.

When you had to scroll through three pages to get what you want. They archived what they wanted.No?
In some way, I felt the same some time ago, so I created http://preciousdigest.com to filter relevant tweets in my timeline.
Where is Twitter getting this money from?
They have a lot of cash on hand: http://finance.yahoo.com/q/ks?s=TWTR

It's likely money left over from the IPO & previous fund raising events.

I guess they've given up on making the text part of their product make a damn bit of sense.
I wondered this too. Twitter's app is broken - numerous complaints by power users that they should be addressing but instead they're busy buying companies?
I think probably team of people who develop and fix bugs on the app aren't the same team that buys companies, so Twitter can probably do both.
Obviously it has the resources to do both. The point I'm making (badly) is that if their core product is broken perhaps buying other companies should be put on hold until they fix it.
Ah, gotcha. When you stated that the app is "broken" I assumed you meant buggy, not "broken" as in a bad core product.
Worked great for Yahoo.
This headline is far better if you don't know what Magic Pony is.
TWITTER ACQUIRES MAGIC PONY; Joke Writes Itself

Come to think of it, this is also more evidence for the case that Silicon Valley can't get any more outlandish than Silicon Valley, isn't it?

I assumed it was something about Hasbro's My Little Pony series then actually found myself disappointed when it wasn't.
Lvl6 Unicorn : $1 Trillion evaluation, cross board agreement on perpetual 0 profit.
I'm assuming Magic Pony is something Twitter will ride upon to bring themselves to profitability.
Even a magic pony can't save Twitter.
That's why Twitter also plans to acquire Magic Beans. Guaranteed to grow instantly up into the sky!
Magic Enterprise Beans?
All I know is now I want a Magic Pony.

Also, was kind of sad we can't post images, this story would be flooded with so many pony memes.

Another sad moment for the London startup scene as one of the most promising startups of recent years get acqui-gobble-hired up by $BIGCO
For most companies in the startup scene getting bought by a larger company that has the scale and resources to apply your tech to bigger and more interesting problems is the planned exit strategy.
While not 100% comparable this feels as weak a move as selling Facebook to Yahoo for $1B

London needs strong, original startups. It's dire if you're looking for employment out here.

Their product was only of use to a bigger company.

Just like the slew of CV startups getting bought by Snapchat.

Tensorflow, MagicPony.

I sense a trend here (machine learning).