Hacker News Reader

Hacker News

Comment by josh2600 | Hacker News Reader

Full thread

josh2600

Where's Rob Spectre at when you need him!!

So many questions.

1) Does Twilio own an MMSC? How did access happen? This is no easy feat.

2) How does signaling work? MMS is an incredibly rich format and can do a ton of stuff SMS can't.

3) I'd love to hear the backstory on how this went down. I imagine this is multiple years of work coming together.

Yeah, I'm a telecom nerd and I know how politically difficult this is to achieve. Incredibly impressed if this is real MMS. That's one of the carrier grails, so getting that exposed is awesome.

Would love to hear more if possible!!

View on HN

Hanging out on the floor at Twiliocon, Josh - you should come!

Great questions - as you can imagine there's not a whole lot I can share about the relationships with our carrier partners. Working with them was essential to bringing this product to market, and we sincerely appreciate all the hours they put with us on getting Picture Messaging out the door.

A few details to the questions I can share:

1) The signaling, the SMIL generation and whole lot of other headaches from the MM* protocols are completely behind the API. We tried to keep this new feature as close to the SMS API you already know and love as possible - for inbound there are just four new params (NumMedia, MediaUrln, MediaContentType and the new unique identifier MessageSid) and for outbound just one new param MediaUrl.

You send / receive your photos - we make it work for whomever is sending or receiving. If you're at Twiliocon this week, two of our messaging leads Thomas Wilsher and Kelvin Law are dropping a killer talk on some of the technical details around this abstraction. It's probably my favorite of our talks at Twiliocon right now.

2) This was the combined effort of a whole lot of talent inside and outside Twilio for a very long time. Very proud of what they've been able to accomplish working together and so stoked to see first hand how excited developers are to use it.

3) This is real MMS sending and receiving pictures. We rolled out Picture Messaging first as we wanted to get it in developers hands as quickly as possible.

Got to head back out and join the Doers at the Design Concourse in San Francisco - Twiliocon only comes once a year, so I don't want to miss more time with all the hackers that came out to hang.

Always great to hear from you Josh. You should swing by if you're in town.

I wanted to go so bad!!! We're over at TC3 today, but I think Shells from our sales team is out there. If you can, you should corner her! She's awesome.

Congratulations on this feature. You and I both know how ridiculously hard this is and I commend you for going after this.

One quick question: is mms a short code only thing or is it available on long-codes as well? Some of the language in your marketing materials left me unclear.

Thanks, and again, Shoutout to you, Jeff and all the other amazing folks at Twilio. MMS access in telecom is my definition of "killing it".

You guys did a great job. Having worked with MM4 and MM7 I can imagine the undertaking Twilio had to take. I'm sure the worst part must have been the MMS provisioning with carriers.

Looking forward to other media type soon :)

Serious question from a non-telecom nerd:

Who benefits from designing a protocol and network for exchanging multimedia messages that makes it "politically difficult" for new entities to participate, and requires years of work plus owning an "MMSC," which you make sound difficult?

From an Internet point of view, it seems obvious that the value of this protocol and network grows with the number of participating entities. We've understood this since the days of fax machines and Metcalf's law.

I assume the competition here is MIME e-mail with picture/video attachments, which is implemented most places and works pretty well and doesn't take years of work.

If you make your own protocol hard for others to join, you're more likely to lose. (Just as the Information Superhighway and OSI protocols lost in the 1990s to the underdog Internet.)

What are the upsides to doing it this way?

Truthfully, my personal track record for predicting protocol success isn't particularly good. There is an old essay on gopher's "resurrection" that I really hope no one is able to find anymore.

All I know is that a whole lot of our customers asked for it, so we built it. Took us a long time, but I'm very stoked for the response developers are giving us at Twiliocon.

We recently launched a product with a couple short codes with you guys. We intended "going direct" as volume and time permitted (we are not telecom nerds...)

My co-founder, hell bent on getting MMS as soon as possible lead the charge. We would meetings where I would repeatedly say "bullshit", and he'd say "no, really... it is the most fubar'd thing you've ever seen"

So a stoked response is an understatement, anyone that has looked into doing this knows you guys hit a homer and I'm looking forward to the wild ride of mobile comm and seeing how push, texting and all that play out. I think MMS available to developers en masse gives it a big push in the right direction and I'm beyond excited that our short codes are hosted with twilio!

The wireless providers in the US do their best to keep SMS a walled garden. They severely restrict what you're allowed to do. In some way this is a good idea as cuts down on spam.

As far as the complex protocol part, I'm too familiar with MMS exactly although I did read the SMPP spec once. I get the feeling the complexity is partially design by committee as well as a bunch of referencing other "standards" by the ITU. For example, for text messages, there's a language identifier for "Pennsylvania Dutch".

The upside to any complex protocol is that it raises the bar to entry. If you can make implementation 10x more difficult, and you're a large company, the extra implementation cost hurts you little, but might prevent a smaller competitor from getting in because they don't handle strange cases. It's hard to tell whether the complexity in protocols comes from this malicious intent, or just incompetence.

Its definitely real! They officially support Images, but will accept (and pass along) lots of other formats: https://www.twilio.com/docs/api/rest/accepted-mime-types

+1 to this. Make it happen twilio.

Reply on news.ycombinator.com