Hacker News new | past | comments | ask | show | jobs | submit login
Should I give up?
49 points by anaphoric on Aug 14, 2008 | hide | past | favorite | 47 comments
Hello All,

I would appreciate very frank advice on what I should do with my start up, Anaphoric Systems (http://www.anaphoric.com).

The whole effort started as a research project several years ago, but when I was informed that I actually owned the system, I entered a business plan competition and then started to execute on the plan.

The problem is that I have been wearing two hats: 'researcher/educator' and 'entrepreneur'. This has led to conflicts of interest as well as diluted effectiveness. So basically I need to decide if I am going to 1.) quit my tenured position, move to a start-up hub, find a partner, get-funding, etc. or 2.) open source the whole project and treat it as a research project. (It's 18k lines of LISP). I suppose in case 2 I could always do something commercial if the system was of high impact.

I am leaning toward option 2, but I thought I should run my plan by the YC community first.

Regards,

Michael Minock (http://www.cs.umu.se/~mjm)




First, kudos to Umeå university for being forward-thinking enough to (gasp) allow you to own your own work.

To answer your question: If I were in your position, I'd try to work an option 1.5 -- see if you can take an unpaid leave of absence for 4 months and spend that time working on your startup. Most universities are happy to let their faculty do this sort of thing, since they usually come back far more excited about their research than when they left.

Spend those 4 months working on your startup full-time, and plan on deciding at the start of the 4th month whether you're going to take your option 1 or your option 2. I don't think you can really make a rational decision without working on this full-time for at least a few months.

Re your startup: Have you considered putting it together with freebase? I don't know how structured your input databases need to be; but since freebase is basically wikipedia (EDIT: and apparently other data sources now, too) parsed and converted into a database, being able to run natural language queries against it could be very interesting...


Actually it's a law in Sweden called the "teacher exception" ('lärarundantagen') and it is not without controversy. But it was the business development folks at Umeå who let me in on the secret and gave me a lot of encouragement.

Yes I am considered taking a leave next Spring. It would be difficult to Swing this Fall. And if I did take a leave, I would definitely move to a start-up hub. I really agree with Paul on the need for that.

As for freebase, yes I had considered that earlier and indeed I should check back. Alas a nice thing about having my system as an open-source system, would be that I could get a masters student to do the integration with freebase as part of a project...


We (StreamFocus - lisp based) are on our final UI upgrade push before launching open beta in the next few weeks. We are missing search right now, and would be very interested to license your system to work with us. Right now we use traditional Mysql tables as well as a mysql triples table, but most of the real work happens in lisp.

Are you interested?


Yes, I am interested.

I like your videos. Very well done. We can discuss via email. mjm@cs.umu.se is probably the best mail to use.


Excellent. You will be hearing from me. (We have to redo all the videos because of the major improvements we've made to the system since then, but thanks)


I would first create a list of firms that have strategic interests in your product area (DB and BI Vendors, DM service companies, Lisp consulting firms/vendors, etc.) You might find a company that is willing to buy the technology and/or offer another profitable arrangement. A joint venture, perhaps.

Option 2 might be your only real option, but even then I would think you might be able to use your work as a springboard into something more lucrative...?


If I do option #2, that is open-source the system under BSD, then I suppose that would give me the option to fork the whole system later under a commercial license if the system started to gain traction. Of course the risk is that others would build up from the earlier version and essentially lock me out. However I would be pleased to have made such an impact...


As long as you get copyright assignments from anyone sending you patches, you retain the rights to change the license whenever you want, so even if you use the GPL, you could release new versions under a proprietary license.

That said, if you really just 'give up' any commercial hopes, a BSD style license is friendlier to companies who may want to do something with your code, and, if they're smart, maybe even contribute something back.


Open sourcing is a mixed bag. The sad fact is that open sourcing will generally limit the number of companies that will consider buying you by some substantial amount. Could probably get the poster a cushy job at google, though...


The idea was some sort of middle road between abandoning any hope of commercial success, and doing it 100% as a classic proprietary software business. I have no idea if it's feasible for his product though.


Please don't give up! I just tried your geography demo, and I'm no expert, but I was very impressed.


Thanks for the encouragement... being impressed with the demo is ultimately what matters.

A serious problem I have however is that this geography database isn't so interesting. Who is really going to learn anything from this rather low quality database? What I am looking for is some database that is: 1.) complex enough to preclude standard forms interfaces and 2.) contains information worth asking about. Not an easy task to identify, obtain such data sources.

I was thinking maybe a sports scores database or something...


I wouldn't discount your product's usefulness for simple datasets. As a UX guy I'd love to see this put toward a better ecommerce search/browse -- "Desk chairs between $150 and $225".

I see it augmenting the standard facet-filter style browse. For example http://nymag.com/restaurants/: I enter "indian food in the east village" -- your product helps pre-arrange the facet filters for me, but I can still add a price range or another neighborhood to the query through standard UI.

The obvious benefit is that natural language browse offers quick access to complex queries (within the user's mental model). But if the results are presented well, your product can be used to train the user to understand the mapping between natural language and executable query (and in that way inform their mental model).

If that means that some users only need to use the natural language feature a few times (before switching to standard UI), it's still successful.


1. I ignored all the samples, and typed:

what's the biggest city in the us?

The response was:

"Sorry, but 'what's the biggest city in the us?' is not understood by the system. The system does not understand what the word ''s' means."

Which was disappointing.

2. Next up:

does portland have a bigger population than seattle?

Response:

Sorry, but 'does portland have a bigger population than seattle?' is not understood by the system. The system does not understand what the word 'bigger' means.

Boo. Also the word 'system' has no meaning to ordinary people, and only vague meaning for the rest of us.

3. Have you considered using a greater dataset for a better demo? Ie, police crimes, or CC data from wikipedia?

4. Show us what you can do on the front page. I don't give a damn about your sepia toned dude writing on a blackboard, or your brown colored optical fibre stock image, or that you, like everyone else, intends to be the prime supplier of whatever the hell it is you do.


Tjena! This sounds like it would be useful for me (haven't tried the demos since the server is down) but I have a pretty big and messy set of astrophysics simulation data that I'd like to be able to query. I doubt you'll get much traction from the public with that as a demo, though... ;-)


baseball statistics could be good.. there are a lot of baseball stats nerds

who had the most stolen bases from 1935 to 1955? how many people have had back to back 5 hit games?


Yes. These are exactly the kind of questions that the system can support. There is a big baseball database out there that I played around with earlier (http://baseball1.com/). It's only for non-commercial activities the last I checked.


You should check with a lawyer, but I am pretty sure that mlb statistics are public domain, just copy the database.


Don't do it. If your system is good enough, there will be demand for it, and you'll KNOW when the time is right to quit. Otherwise, you're taking a needless risk - you can have the same benefits without the risk just by making some clever decisions and delegating properly.


I'm not saying you're wrong...but there are some things about starting a new company that cannot be delegated. Like the strength of your own focus on one single goal.

It may not be the right time for him to make the leap, but there is always a right time to make the leap (or give up on the business or on entrepreneurship in general). Since he's asked us here, and seems to have put a lot of thought into it, I suspect he is now at that point. Thus it is not a "needless risk". It may be exactly the risk that needs to be taken, and it may be exactly the right time to take it.

Entrepreneurship does involve putting some skin in the game. It's not optional and you generally can't be a regular guy with a regular job, and also a big entrepreneurial success. At some point, you have to decide what you want, and go after it with vengeance.


Yes I am at that point... The problem is that I am now finally getting reasonable academic success. A lot of people want to collaborate on papers, projects, etc. And I must say these are some of the best researchers in the field. The only thing I am certain of is that I will be working on this NLI problems for many years to come. The question is which side of the fence will I land on. I can not keep riding the fence for much longer ... or, who knows, maybe I can ;-)


Philip Greenspun might be a good subject of an afternoon of reading for you. He went from academia to startup and back to academia. He's been pretty successful in both fields.

Randy Pausch, of "The Last Lecture" fame, talked about the subject in that lecture. Though he was coming at it from a slightly different perspective: become an Imagineer, which was a lifelong dream, or go back to teaching...he chose teaching. Working for Disney is not quite the same as starting a startup, though I imagine for some folks it would be the same kind of personal satisfaction...that of building something real.

I don't know that they've ever answered specifically why they made the decisions they made, but it couldn't hurt to get a feel for their thoughts.

Two other quick points, before I become a preachy bore:

1. If you're getting success in academia, you will probably also be able to get success in other fields, if you apply the same passion and excitement. The question shouldn't be "what will make me most successful?", it should be "what will make me most satisfied with my life?" As you get older, you'll recognize this more and more clearly...better to spot the trend early. Success will usually follow.

2. Academia, at least in technical fields, does not generally frown on success in the "real world". Success in business can be as good for your academic career as the papers you would have written and the research you would have done during that time. Again, follow your satisfaction and your excitement.

3. (I know I said 2...I discovered this one just now.) You will never be in a better position to take risks than you are now. As you age you collect responsibilities: Family, kids, parents needing care, mortgage payments, etc. These responsibilities are the things you can't risk to pursue your dreams...so, when you're older, you can't put as much in the game. You can still start a startup, or change careers, at any age, and experience certainly has benefits, but when it comes to risk, younger is better. Without family and kids, the risk is practically nil.


Those very people who want to collaborate with you are the ones who will send you customers. Use your university to drive customers to your product - when you go to silicon valley, you will be just another no-name with a big idea and some code. You have the name and rep right now, create your business out of that, and when your business is big enough, go to silicon valley.


I don't know the details of the system you've created, but you could use a 'restrictive' open source license like the GPL, and offer a dual licensing strategy, where people who want to use it in commercial systems pay to get a compatible license. That only works if it's a library though, rather than an independent system.

Another idea might be to find a partner first, and let that person handle most of the business, and just concentrate on the technical aspects of it.

Good luck!


The system calls several GPL based components (SPASS, aspell, ...) but does not extend any thing under GPL.

I was thinking that if I open-source it I would put it under a very open license (e.g. BSD) and assume/hope that anyone that extended it would make contact with me for help/advice.


I know this quandary well though I was a post-doc when I took the plunge.

Why does it have to be #1 or the other? Could you see about taking a one-year leave of absence from your tenured position? Seems that if you're successful you'll generate a lot of good publicity for your university. Furthermore, you could try to find a visiting relationship with a university at one of the hubs. For you it would be a good way to meet people and your current university could see you're still serious about the academy even if you only "visit" a minority of your time.


We can't answer that for you... but we can tell you a bit about our own situation.

Personally, I've been running my own (much simpler) VPS hosting startup (http://prgmr.com) for more than two years. I'm on my second Porsche worth of personal money plowed into the thing (ok, second Porsche boxter... I'm probably still on my first, you know, real Porsche.)

I dono. I mean, my company breaks even these days... most of what I've gotten out of it is experience and contacts (and those are worth something... my bill rate as a consultant verges on ridiculous, and I have a book deal. Pretty good for a Midwestern hick with no formal education.) I don't regret doing it, but if I had a chance to do it over I'd focus more on being more public (blogging, documenting, etc...) and less on the actual business.

If I could 'open source' my project, I think that would magnify the benefits (that is, credibility and contacts) But you can't really 'open source' an ISP, which is essentially what I am.

So yeah, personally I'd advise you to go for option #2, unless you can get significant investment from someone else.

But then, you sound like you started from a different position than I did. It's possible that because I had so little when I started the credibility is worth a lot more to me than it is worth to you.

Also, I have my own conflict of interest... your project sounds pretty cool, and I'd like to be able to use that technology for my own projects, or use products built by other people using that technology. Open-sourcing it is likely the best way to get the tech out to the masses.


How close are you to getting sabbatical? (Does your university even offer it?) Several professors in my department (including my adviser) have used their sabbatical on startups.


Unfortunately sabbaticals aren't part of the Swedish system... Still I am sure I could swing a non-paid leave starting in January or so. Perhaps even in November. But not in September/October! I am already down for teaching AI.


Michael, Neither of the options you have provided would equate to 'giving up'. So I would give a definate NO to giving up.

Treating your project as a research project does not equate to giving up.

I don't know enough about this market/competitors/technology/market size, etc to define what you should do going forward.

The question I think you should be asking is 'is this a viable business'.

I suspect most of the YC community who are telling you to go for it! are potentially speaking from a start-up attitude of 'live the dream' etc. etc. They may be right, but not every business is a good business.

Your technology might be the best out there, I just don't know. Personally, I wasn't blown-away by the demos, but I'm not saying that to be harsh, just that from my perspective, I'm not seeing your competitive advantage. You've put a bunch of work into this, and it may be the best in the market and able to make a significant contribution to the world.

Does your university have a business school, or entrepreneurs club or something? Maybe you could get somebody there to help with defining the market and business planning.


Not wanting to be a wet blanket on this party but this system has a looong way to go. Too many of the queries I made either gave me dumb error messages or decided to answer a different question.

Try and search for the shortest river. No dice. Given that it knows about second, third and the like I tried to rephrase this as "last longest river". All I got back was the longest river, it dropped 'last'. It knows average but not mean. It reports 'average length of a river' as 1117.2391304347826087. That's a great deal of precision for something without a unit of measure. What is this in: meters, kilometres, centimetres, feet, inches, miles? Given that the data is from America I would expect this to be in miles but I suspect that it is probably in kilometres.

Has this really taken 18k lines of code?

Stay put and knock this thing into shape on the universities dime then start your own company.


Hi Peter,

> this system has a looong way to go

yes it does... Natural language interfaces are a very hard area. Lots of failed attempts (e.g. MS EnglishQuery). This in fact is what attracted me to the area. If as a researcher I can help make NLIs more robust/usable, then that would be a 'result'.

> Has this really taken 18k lines of code?

Actually that is an approximate figure and includes test routines, comments, etc. Plus what is a line of lisp code? A large amount of code is to support non-linguist authoring of the interface. So in fact I was able to patch the system for "shortest river" and "mean" in a jiffy.

> Stay put and knock this thing into shape

Yes that's what I will probably do. I toned down Anaphoric's web-site and will work with this system from more of a research perspective. I will try my best to avoid open sourcing it, but in the end circumstances may compel that move...


How certain are you about your presumed choices?

You ask us to choose either 1 or 2, yet we have no idea what situation and/or persons brought you to this apparent fork in the road. I suspect there may be a 3rd, 4th, &etc paths that prove better. Nor do I see 1 and 2 as mutually exclusive as you state. Many, if not most, researcher/educators are simultaneously entrepreneurs. And giving up tenure may prove costly, especially in the long run.

Perhaps tell us more about the conflicts of interest that have arisen. I sense that they involve other peoples wishes or meddling, which are usually best ignored or stepped on.


one todo, I think, clarify your website for the obvious questions, so people don't have to download the demo and watch screencasts to get something out of it:

- what is the benefit of your algorithm and code over Lucene/SOLR? quantifiable metrics, e.g.precision/recall-wise? What natural languages does it parse/work best in?

- what kind of analysis, in layman's terms (Layman: somebody who's implmeented Lucene, sphinx, the search engine, not CMU sphinx, MySQL fulltext search, not an NLP researcher) From the company name, i would assume anaphor resolution is included, or something related, named entity resolution.


Very nice design. But:

"Server Down on localhost:9009 Please visit again in a couple of hours."


OK, yes, someone's query busted the system down. But it's up again now.


Michael,

My suggestion is that you use your salary to hire someone to work fulltime on the project. You are presumably in a good position to find a smart employee who is interested in your work and will be able to contribute within a short period of time.

If you cannot make headway in the market with two developers/proponents, it seems unlikely you will succeed alone. Not all of the work needs to be technical - there's no reason you can't hire someone and have them handle marketing/sales.


This is very interesting technology something our startup like many others could use it. Server licensing is always tricky. I would say go with number 2 and in the meantime build a commercial product around the technology. Since you understand this technology the best it wouldn't be hard for you to build a shopping engine based on CatchPhrase.


This is neat stuff.

Actually we (reviewgist.com) could also do with something like this. Infact, we spent some time working on a similar system for our needs, but anticipating a dis-proportionate consumer response dropped it. But if you go through our site, you would know exactly what I am talking about.

Also to answer your question, imo go with 1.


I tried a few queries and I was really impressed. Seems to me like some real value for the IP there and wouldn't recommend going opensource unless you already patented it(Europe but I think doesnt have software patents right?)


This is really neat. Especially the idea of allowing users to modify the database. How would one add data to the demo to indicate that Cambridge, MA and Silicon Valley are the best places to start a tech startup?


IMHO option 1 is the way to go if you don't have a family


Well I am divorced and don't have kids, so the world is my oyster ;-)


I'd go for option 1. It seems like a solid system and it does have value. I would love for my city's library system to use it for example.


Step 1: Do #1 as you mention

Step 2: Quit and start your startup

Step 3: If things don't work out, then go to your step #2, and move onto other jobs


Michael, I suggest this:

1. Don't spend any huge money on this project until it picks up significantly.

2. treat this as a hobby, which means don't lose sleep over it.

3. focus on your work that makes money if money is a factor

4. take it easy. leave the company as it is.

5. try to make this not be written in LISP. that is a lot of lines of code. convert it.

6. Keep everything going the same time. don't quit your job.


Wow, this is horrible advice.

1. You have to take risks sometimes. He's done good work and it's worth putting some cash behind it (if necessary).

2. It's not a hobby. It's a business or a research project. In either case, it's a serious thing.

3. This project could be a money-maker in two ways: corporate sponsors for a research project, or corporate clients for a business product.

4. I agree with the take it easy part, but not the other part.

5. That's a joke right?

6. That's also a joke right?




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

Search: