I have invented a new way of filtering email that is so accurate it’s scary. I’ve been in the spam filtering business (Junk Email Filter) for 15 years and if you asked me 2 months ago if this level of accuracy is possible I would have said NO WAY! This thing is almost 100% accurate, especially on not blocking good email. I’m calling it the Evolution Filter because it seems to mimic the way evolution works.
I’m thinking about getting a method patent on it and I’m torn on the issue and need advice. I’m in the process of writing and filing a provisional patent so I’m not going to tell you how I’m doing it today.
I can tell you for sure that no one else is doing this because if anyone knew this trick, everyone would be using it. It does have a learning system. It does have a database. It is not Bayesian or remotely similar to Bayesian.
So you all know me and you know I’m not just making this stuff up. So for the purpose of this discussion let’s assume I actually am onto something.
I’m about to file a provisional patent. The idea here is to make it good enough that it will be east to convert it into a real patent within a year. It will also create the illusion of protection while I’m trying to make a buck off this. My fear is once I let the cat out of the bag it might put me out of business. This process is so accurate that if it were widely implemented spammers might just give up. It’s that good. And – although the method isn’t obvious the implementation is trivial. I had it working in a couple days and my programming skills are good – but not outstanding.
I have been advised that I should just keep it secret and just have the world’s best spam filter, but that’s not what I want to do. I want spam to die. So what I would like to do is release it for free to the public but be able to charge the big guys a licensing fee. Something like 1 cent per email account per year or 5 cents per email account for a perpetual license.
My feeling on this is that once I release it the spam filtering community will go wild. This is going to spread really really fast. And the big guys will notice it very quickly as well.
Having used so much public domain software and methods I do feel like I want to give back to the community. But if the world gave me a billion dollars (not expecting that) for this method the world would be getting a great deal.
I do have some concerns about the method and what else it can be used for. It’s uses are not limited to spam but can be used by any system to identify almost anything.
It is also a very simple recursive learner that is extremely fast. If a server connects to my server and sends 2 emails on the same connection, if the first email is learned the score for the second email will have already been altered.
OK – my question. Assuming i’m not delusional – I’d like ideas about patents in general and ways to make money on ideas without a patent.
Thoughts?
You are using patents wrong.
You don’t invent something new and then patent it. That’s so 1930s.
You invent something that already exists, give it a fancy name and then after being granted the patent you have a lawyer send threatening letters to BestBuy, Sears, AutoZone, Barnes and Noble claiming they are violating the patent.
Does your invention leverage Google’s search engine and analytics?
No – it’s my own learner that is exceptionally simple.
“But if the world gave me a billion dollars (not expecting that) …”
Advice: Buy a PowerBall lottery ticket tonight.
I have a distribution list of friends with 25 or 30 names on it who I send out funny emails to on an ad hoc basis. The email broadcast goes out blind copied to all. Sometimes I add a few names who aren’t on the distribution list, and then there might be 40 recipients blind copied. Everyone is opted in.
What would your spam filter ‘think’ about my broadcasts?
Depends on what’s in the email.
Almost everything I send is an attached (or embedded) image. Usually (over 90%) there is no text at all in the body of the email. Half the time the subject is substantially descriptive of the attached (or embedded) image, the other half of the time the subject may be a twisted or sardonic pun based on the attachment.
So now what would your spam filter do?
Give me your subject and I’ll test it.
How about I add you to the list until you say to take you off? As admin, you have my email address. If you want me to do this, send me an email indicating such. This would be a good test for you since I believe lots of people share stuff with their friends using similar broadcasts.
Mark, is your filter based on e-mail content, sender, or both?
Why not put up some addresses where the filter is being actively used, and let readers send it spam, as well as legitimate mail.
If you could somehow set up a response automatically upon receipt and passing the filter, then we could test it for you.
How have you tested it?
Couldn’t you just present it to Elon Musk and have him make his own email system?
I’m filtering about 5000 domains right now using it.
Yes but have you done any explicit testing?
Is it just a matter of going thru the spam filters and seeing the accuracy rates?
I wonder about what happens after your method is known, with the spammers knowing what they are up against, what type of spams they might make to get thru the filter.
What’s different about my filter is that the more they misspell stuff the easier they are to catch.
Also – Bayesian poisoning doesn’t work against my system.
Bayesian poisoning is nasty stuff.
My poor little pooch walked through a puddle of bayesian fluid the mechanics spilled, then licks his paws.
Great Marc, you discovered an antidote for Bayesian poisoning, amazing. Are you selling bottles on ebay?
The taxpayers of the USA could not afford that.
Quantify ‘accurate’
One more thing, how does this compare to Yahoo’s and Google’s systems? I assume MS Hotmail doesn’t have any spam filter.
I find Yahoo to do a good job, and I barely get any spam.
Accurate would be 99.9% catching spam. 99.999% not blocking good email.
Do you assign a score to spam? So that if I wanted to check what got filtered out, I could see “These are the most likely legit e-mails filtered out by accident.”
One of the best features of the system is that it actively detects ham. It is very good at identifying good emails in a way that spammers can’t imitate.
So I’m not just looking for spam, I’m looking for ham.
What do you do with e-mails that are coming in with the unsubscribe option, but probably not requested by recipient?
Excellent question. Such emails are legit unless/until the sender ignores any unsubscribe request.
Similar to my funny stuff broadcast situation, it could generate a false positive for spam.
Marc?
Build and test it first – no need to tell anyone about it. If it works, then patent it. Once it’s been patented, then advertise/sell/license it as much as you want. You only have an idea at this point. Ideas are a dime a dozen.
It’s built and running now.
And how does he keep people from stealing the patent?
What I want to do is license it to the big players for a very reasonable fee. A patent would give me enough ownership to make them want to pay me rather than risking losing to me in court.
When I patented something a while back, we hired a lawyer to write the application. That was TOTALLY worth the thousands of dollars it cost, but it required that we explicitly spell out the process for the filing.
On the down side: a patent is only as strong as your willingness to litigate in its protection, and that will be expensive.
take a look here..
https://en.wikipedia.org/wiki/Alice_Corp._v._CLS_Bank_Int'l
now kiss your dreams goodbye 🙂 we are all the better for it
I’m familiar with that case. It’s not a software patent. It’s a method patent.
Don’t patent it. License it. Worry more about making some money than spending tens of thousands burying your idea in the us patent office.
But I want to be able to give it away for free to the open source community.
WHO THE FUCK (in their right mind) USES EMAIL?!
Nevertheless, good luck with that. Though I do find it a bit funny that someone devoted to a “religion of reality” is trying to patent a fictitious device or place a trademark on something that will never be real enough to touch. Hell! If Bill Gates and Steve Jobs can do it…
I have it running now. It’s filtering Dvorak’s email.
So that’s why he never answers my emails.
And I thought he didn’t like me!! 😉
Don’t take it personally. He ignores it when I point out that his home page (text only) is full of dead links. I still use it tho as my home page (it is a text file on my computer connected to the ‘home’ button on my browser) to select a browser for every new search to break the chain of surfing collected by the next site I’m visiting.
see http://dvorak.org/home.htm
Go for it.
I considered a design patent for a fish tank top terrarium design unique at the time. I did my patent research and saw that immediately upon patent filing it would be ripped off by the Big Merchants (Matel, Price Fisher etc) by changing something simply and obvious claiming a new idea…like changing a hinge mechanism to a snap it one.
So, to your question, if you want to make money on something that is easily stolen, you have to have immediate market dominance by skipping the patent and going straight to the market. I couldn’t do that. I used my model for about 10 years and then tired of the hobby. Still haven’t seen anything like it.
So………the community lost out to the lack of patent protection.
I watch Shark Tank all the time. Mark Cuban is big on making software deals, so is the Robert guy, and Mr Wonderful claims expertise in licensing deals. There must be licensing gurus all over Silicon Valley? A few phone calls or emails should get you some direct advice????
I see lots of software available for free to the general public or home use but fee for service to the commercial world.
What about working with some server outfit who wants to stop the unnecessary work load of moving spam?
Sell out to the NSA?
Hey Marc—you strike me as a happy guy. don’t lose that focus in your life, whatever you do next.
An informal “trick” done for copyright works is to send yourself several registered mails with your work product opened on the inside. This establishes an evidentiary date of your work product. Not as strong as any actual filing, but it is totally secret…. except the USPO is scanning every piece of mail…. I’m sure for mega data only. I’d check that trick if you want to rely on it for anything except your personal benefit.
Again………..keep it lighthearted, and have fun with it.
UNopened on the inside ……see what happens when you let spell checker do its thing? Fix THAT next!
I don’t know software, the internets, or IP much at all….. but could you start your own “Filter Service” by leasing then buying your own service and having folks route their accounts through you?
I assume this would keep your software a secret…. or would any such service even on your own server be immediately hackable?
Assuming it is….. looks to me like partnering with the biggest player in the business, or maybe with No 2 who is trying harder, would be the play. Enough to buy a Tesla….and drive with a smile?
Even for someone who is weak in reading comprehension, that is horrible. I take it you never read anything Marc writes?
NFS–if you are talking to me, what do you find so horrible?
Educate me/the rest of us.
It’d be entertaining to watch Marc on Shark Tank.
I imagined if you started now, you could make it this season.
This isn’t a design patent. it’s a method patent. It’s also not a software patent.
Another quality Post.
‘I had a really great idea, look at how great i am , but i won’t tell you what my idea is’
I had a an idea for a blog by John C Dvorak , but without some failure posting his garbage all over it.
Stay Worthless
So where do you draw the line on “good email” versus spam?
Have you ever read an unsolicited passive email (no virus) that turned out to be interesting or useful?
Anyone?
If I told you where the line was I’d be giving it away.
Ahh… so the question is really: what (simple?) characteristic does all spam have, that normal emails normally don’t have.
Maybe header analysis of third hop softfails and the presence of prefix (SRS0)?
MARC!
I’ve been doing patents for the last couple decades and make a living doing product development.
Allow me to make a few suggestions (many of which I’m sure you have already considered):
1) Patenting the method is the right way to go, starting with a provisional patent. I’ve done 40+ patents over the last couple decades and had only one turned down.
SO LISTEN TO THIS ADVICE: DON’T DO IT YOURSELF! Wait, wait! I’m not saying you can’t do a great patent, you probably can do a better patent than most lawyers.
What I’m saying is, if you are going to defend or sell your patent, nobody gives a damn how good you are. If google want to use your method, a well known law firm will stop them in their tracks because they know you’re serious and they could loose.
If someone wants to buy your patent, they can justify the cost to their bean counters and investors by pointing out it’s a quality patent because it’s backed by so-and-so law firm, and it will be cheaper to buy than steal.
A good law firm will be expensive, expect $20,000 if you do most of the work yourself (and they then rewrite). BUT – and I’m serious – a good law firm on your patent is worth an extra couple MILLION dollars. That’s 100x your investment, and I’m being conservative for what you have.
If I may, let me offer you additional considerations which relate to what I have already said above:
1) DON’T GIVE IT AWAY. This isn’t something like an antivirus or shareware that your grandma can use, it’s a serious back-end product for professionals, and therefore only useful to serious pros that can install and manage it. And they have to buy it.
2) Related to 1). You live close enough to Google, Yahoo, Facebook, etc. to shop your product. Most may find it hard to get in the door but you’re not a Johnny-come-lately so you will get in. Setup a test account where Google/Yahoo/Microsoft/etc. can forward test emails so they can monitor the results. Sell them on the benefits of a better spam filter.
Yahoo and Microsoft sure have a spam problem, and Google buys everything in sight.
3) Million dollar idea: Facebook. Facebook has a huge bullying/demeaning problem. It’s so bad I had my daughters leave Facebook. Show Facebook how your method could manage and control these postings. Show them an opt-in “friendly neighbor” policy where bad language, bullying, etc. posts never gets displayed. Too many bullying or abusive IM’s and timeline posts gets the offender blocked. This is a no-brainer for them. It will help the community and their reputation (and hence their share price. A 5% improvement in share price is worth hundreds of millions, making your product free to them. They actually make money using your product!). If Facebook will go out of it’s way to offer 50 shades of whatever-sexual-orientation in their profile, they will pay dearly for a way to effectively moderate hate speech. Now I’m not talking about an ad blocker, but a hate speech/abusive/bad language/etc. blocker specifically for postings and Im’s to improve the overall experience.
I know for a fact Facebook has the means to pay millions for something like this. I have a friend who sold Facebook and Google some hardware to marginally improve their networks. Now he has enough money to buy his own island. Not joke.
4) Don’t forget twitter either.
5) MOST IMPORTANTLY! Can I borrow some money?? 🙂
Thanks for your good advice. I have some legal background too and I know what you are talking about. I’m doing the provisional to get the date saved and start selling it. I do know that the drafting of the claims is the trickiest part.
I think my provisional is written very well and it will make the work a lot easier for a real patent attorney. I’m going to a startup group tomorrow to talk to investor type and often there are IP lawyers there too.
If I do a full patent I will probably hire an attorney. We will see how this goes.
I am planning to release it after the provisional is filed. I think the world will go nuts over it and everyone will be doing it. then the interest in my provisional will be very high.
I’m finishing up the abstract today. It’s about 13 pages long and it talks about everything. There are actually 3 separate unique tricks I’m doing, the main one I’m sure no one is doing, but the two other tricks with it might be novel too.
Outstanding! Good luck. If I can offer any additional help just ask (email?) if you want.
Marc has experience with patents, having previously patented solar powered motion enabled bicycle safety light.
What was the result of that Marc?
That was sort of self education at the time. I just wanted to see if I understood the process. The year expired and nothing came of it.
Bottom line: Regardless of how good your product and patent is, unless you are a large company with a deep bankroll, it is still essential worthless. The cost of defense – and it is almost a guarantee that your patent will be infringed – is too prohibitive for most all but large corporations…
If the patent is good and the license is reasonable and the value of the method is there I would think they would pay me a license.
I think you are being naive about that. If the method is as easy as you describe, then it could be implemented in what, 1000 man hours? Even 100,000 man hours would save money to copy.
And presumably it is much less than that, since they would still have to implement it themselves.
Marc’s right. Yes it is expensive but the price of reproducing his work plus litigating against him plus the bad press plus the possibility of loosing is expensive also.
Fighting a patent lawsuit isn’t as hard as many make it out to be if you are the little guy.
Reason is, if you have a solid patent (hence having a reputable firm do it) and if your product is worth a few million (as Marc’s should be)-
Then there are numerous hungry patent attorneys who will gladly take on the lawsuit Pro Bono for a piece of the pie.
This isn’t the 1960’s where car makers tried to steal with impunity (and got caught).
The big companies know this. The bigger they are the bigger the payoff.
Yeah – if I do the patent right and the infringer has deep pockets there are plenty of lawyers who would sue for a piece of the action.
Sometimes – and I’ve seen this – it’s more profitable to sue for infringement than actually produce the product itself. That’s the entire concept behind Patent Trolls.
It also works for legitimate “for real” patent holders.
Oh, and do us a favor and keep us updated! This promises to be a grand wild ride.
They have to reproduce his work in either case.
The issue is how would you know they had stolen your patent?
The provisional patent is filed. I’m going to talk to some investor types tomorrow. I’m hoping to publish the details about how this works soon.
Wow the skeptics are pretty brutal. I guess that’s a good thing.
I want to know more…
I don’t blame them. I would be skeptical too.
I am all for anything that alleviates world hunger and suffering!
I’d even go Door-to-Door to sell bayesian fluid clysters if it ends malnutrition in South Sudan!
Dear Marc,
Love your struggle to keep on improving the filter and technology (read the spamassassin mailing list)
My suggestion is: do not ask patent as it is expensive, and you can’t afford big money to defend when others steal technology from it (also spammers might study the information you want to patent).
My suggestion: a price model like Spamhause has for it’s blacklists. And you will earn money as from e.g. more than 100 addresses, in that way you can also invest in hardware and defense against DDOS in case this is getting very successful!
Warm regards,
Tsjêbbe
I’m no expert (AT ALL) but this strikes me as an excellent suggestion.
Subtle.
Protected.
Long term service.
I almost went ahead and filed my own patent and not care if it got stolen or not…. just to be a part of the Patent Files “forever.” A sense of satisfaction in having something file-able at all.
All does depend on what Marc’s central issue really is. Being the author/innovator of something, or making money. Different issues/personalities for either action…. never disallowing for a mix of either and something else as well.
Good luck, Marc! I hope it is a success!
I think I know what the method is. Instead of a black list or white list, you use a gray list. 🙂
Either that or the black and white lists battle it out, and somehow a peace list works out in the end.
I was using my ISP’s Spam filler, for my email account. Until they censored a letter from my sister about a video link on Youtube about CA utility meters catching fire. They never blocked any other YT links, in the past. So I figured they were protecting the Utility Co.s’ interests, not mine. So I turned that spam filter off, and use Thunderbird’s instead. Never seen any spam, either way.
That’s not it. I’ll drop one hint today.
Everyone else uses ideas based on matching something which means they are matching to finite sets of data. That’s why spammers misspell stuff so they don’t match.
What I’m doing is NOT matching which is like matching to an infinite set of data. And the tricks spammers use to get around other filters makes them easier to detect using my filter.
But why wouldn’t they come up with other tricks to get around your filter?
Or are you anticipating a two-stage process?
My filter works by not matching so the tricks they use just create new spam fingerprints. What gets you around other filters is what gets you caught on my filter.
Do you having the recipients identify things in the inbox as spam, and things in spam box as not spam, and the system learns from that?
I’ve guessed it.
He bought a Tesla and runs the SPAM through the car and it magically gets filtered. Elon is pissed because he didn’t think of it
first.
🙂
Nuttin’ but love brother.
A closer answer is that I created the AI that Elon is afraid of that is going to destroy humanity when my spam filter concludes that the best way to stop spam is to kill all humans.
PC Magazine but Elon Musk as one of the things that will die in 2016.
I’m ready for my Robot Overlords!
Marc,
Having several patents- I’ve been disappointed with both the time (some instances 4 years!) and money its taken to get an inkling of a return on my investment.
I know fully understand your approach– buy my guess is it might be a process. Processes are particularly conducive to doing a copyright.
Copyrights historically have been the domain of writers, artists– but now I have heard of petroleum companies and technologists using this method to protect their works because it costs less and is much faster to obtain, plus has the benefit of being able to enforce almost immediately.
Best of luck, and I hope this might be useful to you.
So now that the provision is filed are you going to tell us about it or what? Are you going to let us end this Friday on a cliffhanger? 😀
Nah, go on Shark Tank, if you come up with a more interesting presentation than that global warming one…
He is already able to answer Kevin’s question,”Why do I need you, why can’t I just start my own company, Mr Wonderful’s NoSpam Email?”
I would consult a patent attorney since you could box yourself into a corner that someone else could provide in a similar fashion without violating your patent.
I’ve learned that being too specific about your patent can harm you in your ability to protect your invention.
In hindsight, I would not want to do this sort of thing alone (unless I was a legalese expert).
Best of luck!
In light of the alice ruling, software patents are much more complicated than they were a year or two ago.
If you explain your method in strightforward business terms, does that give away all of the trick? If so, its probably not patentable anymore.
Why even go for a patent? Just give it away. Any money you make from the big guys, Elon will scam yo into giving it all to him.
Here is a summary of my skeptical points:
You are being overly excited(I imagine readers here are shocked by this) about something that happens to be working well, but spammers will adjust and get through your filter as well.
Big players will steal this patent. You say they would rather pay, but I doubt it.
How would you know if anyone is stealing your patent?
I know small players are free under your plans, but what stops me from implementing it, starting my own e-mail service, and then continue to not pay you as I get millions of customers?
Also skeptical about the level of testing.
If you really want spam to die, and it’s simple for any one to implement on anyone email network, you should open source it. Otherwise if you keep it proprietary you may end up like how AdBlock ended up. Where they get paid to allow ads (or spam in your case) to get though the filters.