guestbook messages archive
Tim Hutton
I had to take the guestbook down, since the volume of spam was
enormous. Here are the messages, for reference. Thanks for all the
kind comments in the last year and a bit.
--------
10/09/2005 - tim: Hello Organic Builder users! Hope the builder is
fun and easy to use.
--------
20/09/2005 - Siamak: Hi Tim! This is really cool!! Could it be a bit
faster, with a larger number of atoms? The way you've implemented
""dragging"" is also nice.
Tim: Hi Siamak. Thanks for your kind words. I may well make a small
atom, high speed level, I think this would be fun to explore.
--------
20/09/2005 - Siamak: You can already make a Turing machine out of
these, as well as all kinds of interesting structures. But, in real
world, you cannot chose the way atoms make reaction. It would be nice
to have few basic types of atoms, with fixed reaction state machines,
and then you could put some of them together to make molecules. Then
it'd also be nice to be able to have a library, so you could use the
molecules already in your library to make more complicated molecules.
--------
21/09/2005 - Siamak: Btw, it would be nice that the atoms could also
separate from each other, according to their states.
--------
21/09/2005 - Siamak: . . which is already possible!
--------
21/09/2005 - Siamak: Amazing! This is the way we'll produce our things
in a near future.
--------
22/09/2005 - Tim: Sorry that level 7 wasn't evaluating, all fixed now.
If you find any bugs in the evaluation code (this is the trickiest
bit) then let me know! I see that level 15 has been solved several
times already - this is very impressive. (I'll have to think of some
harder levels.) Can you solve 15 in such a way that you never need to
drag atoms carefully into place?
--------
22/09/2005 - Ralph hartley: Any chance of getting source code? I would
like to be able to add levels and make other minor changes. A fairly
tricky problem is to do 15 in such a way that it (A) is always
guaranteed to work and (B) is as fast as posible (with no dragging)
and/or (C) uses as few states and/or rules as possible. Another
problem is to do exponential replication (each copy copies itself).
What makes it hard is that you need to prevent two replicating chains
from reacting with each other. I'm not sure it is possible. I *think*
it is iff the starting position contains exactly one chain. (I haven't
done it)
Tim: Email me to get the source. (I'd make it public but would prefer
to work with people on making fun levels available here.) (A) *is*
possible without dragging (have you done it?) while of course (B) and
(C) are hard to prove. And for exponential rep, I don't think you
can avoid the interfering problem with a 2-reactant system (without
using membranes, which is another story...) but I'd love to hear your
thoughts. [Historical note: the source code was later released, under
the GPL.]
--------
22/09/2005 - Siamak: For 15, my solution has 13 rules, and 11 states
for each atom. It replicates in 15-30 minutes on my computer. For
exponential replication, it has the interfering problem, but I think
it is not difficult to remove that problem: after application of each
rule x+y->xy that may cause two replicating sequence react with each
other, we can send some signals to make sure that the molecule still
has a valid structure. If it doesn't have, we apply xy->x+y.
--------
23/09/2005 - Tim: New features (as requested!): (a) you can now change
the speed, and (b) you can now choose between few large atoms and many
small ones. (You may need to shift-reload to update the applet.)
--------
23/09/2005 - Ralph Hartley: we can send some signals to make sure that
the molecule still has a valid structure I don't think so. Whatever
signals we send, the other molecule can send also. After all, it may
be identical to us. The probability of being fooled may be small, but
my rule (A) says it must be zero.
--------
24/09/2005 - Siamak: Ralph, You might be right, but I'm not sure what
exactly you mean by your rule (A). I take the interpretation that all
faulty reactions are reversible, which means if two replicating
molecule react with eath other, it is possible that they detach and
continue their job. Is it what you meant? Tim, Thank you for the new
features.
--------
24/09/2005 - Siamak: Maybe to discuss about the interference problem,
it's better to consider models in which - when a sequence is
completely paired, it remains that way, and the strands don't separate
to replicate more. But - the environment may contain several
sequences at the begining. Now, we can formulate the question
regarding Ralph's requirement (A) as, if completely paired sequences
are the only possible stable molecules that may appear from that
initial soup.
--------
06/10/2005 - Ralph Hartley: all faulty reactions are reversible, which
means if two replicating molecule react with eath other, it is
possible that they detach and continue their job It is not sufficient
for it to be possible, it must be inevitable (probability 1). I think
a sufficient set of conditions is: (1) There is a bound on the length
of any sequnce of ""forward"" reactions (good or bad) starting with a
bad reaction. (2) All reactions that can folow a bad reaction
(including itself) must be reversable. If *all* reactions are
reversable, then there are reqirement on the reaction rates in order
to ensure that the reaction goes forward on average. I am now
convinced that it is possible, but it depends on there being an
absolute limit on how many atoms can fit within a bond l- -city of
the rules... Also, testing for transitivity in every collision would
be costly as it will in the worst case require an exhaustive search in
the whole molecule (when the relation is false).
--------
06/10/2005 - Laurent: it is not costly at all (no search thing) if you
maintain a number corresponding to each string. On a collision, you
only have to test if the numbers are the sames. But ok, that would
not be chemically plausible, but it would make the whole full
interestingly more powerful. Especially for multi-replication, which
idea is central for me.
--------
31/10/2005 - Tim: Hello everyone, It looks like activity has dropped
off a bit here, I guess this is only natural. There is some discussion
going on between some people that have emailed me - just send a mail
to join in. I will set up a mailing list for this, if there is enough
interest. Thoughts? In other news, there's a workshop on artificial
chemistries happening in Bristol, UK in April 2006. It will be a
friendly forum for discussing ideas, so feel free to come along or
even submit a paper. Depending on how much time we have, I may
organise a special session on the Organic Builder and on the rulesets
created by its users, since there have been some really good ideas.
The website is here:
http://www.cs.ucl.ac.uk/research/bioscience/aisb06/ [Historical note:
this workshop was cancelled due to insufficient submissions.]
--------
22/12/2005 - Paras Chopra: Hello, One question. While attempting
challenges, do I have to write all the reactions in one go Or can I
write first set of reactions and then pause and then write second set
of reaction and then pause and this way I solve the challenge.. Is it
allowable.. Regards, Paras Chopra
--------
07/01/2006 - Olivier: Question about levels 14/15. I can create the
ladder, but I can't make sure it will be created properly every time.
If you look at the original molecule upright, it has a left and right
side. I bound atoms to it, starting by the 'e' end and following in
sequence. Then I bound those additional atoms together to create the
2nd branch of the ladder. My problem is that when an atom pairs with
one of the original molecule, it may connect on the left side or right
side. So the atoms that are supposed to become the second branch of
the ladder are not necessarily on the same side, and then can't
properly create the 2nd branch of the ladder. Can anybody give me a
hint how to fix that? I've tried to allow the connection of several
free atoms to the same atome of the original molecule (so that
eventually one will be on the correct side) but then I have another
problem that I may have 2 atoms connecting on the same side and then
the second branch of my molecule my contain some duplicates (i.e if
the original is e-a-f, i may end up with a copy being e-a-a-f). In
all cases I was able to ensure that at the end of the copy/separation
process the original molecule is back to it's original state so it can
try again and given enough time and free molecules it will eventually
create a good copy. But that doesn't seem the most efficent solution.
Any help is appreciated. Thanks.
---------
09/01/2006 - Olivier: Paras, from my experience it seems you have to
have all the reactions at the beginning. The interface will allow you
to add reactions in the middle of the run, but then when you click
'evaluate' you will get a message asking you to reset the simulation
and start again.
--------
18/01/2006 - Tim: Olivier - regarding how to make ladders work. It's
a tricky problem, that. Think about how to set things up so it doesn't
matter which side the new atom bonds from. Everyone - sorry about the
spam here recently, I've turned on message confirmation to get rid of
it. Other news - Level 19 has been solved by someone! This is
seriously impressive work.
-------
25/01/2006 - Ethan: I am very interested and may contribute in his
developpement. But it seems that there is no documentation nor clear
organization. Tim are you the only in charge of it? Do you have UML
diagram or a kind of specification sheet ? The only thing I have is
mainly the code. Is there a great diffrence between organic builder
and what you call squirm3, which? What is the latest code? Thanks :)
Tim: Hi Ethan, Thanks for your interest. I'd love to see your ideas -
email the mailing list or me for detailed answers. There is no UML
diagram or suchlike for the code, I never bothered - sorry for this.
:) Feel free to rewrite it... There is not any significant difference
between OrganicBuilder and Squirm3. "Squirm3" usually refers to the
idea of using a simple artificial chemistry in the style of those in
the papers, and the idea has been developing. The code for the
OrganicBuilder is for a continuous-space version of squirm3 (plus java
interface bits and challenge levels). The underlying chemistry is the
same whether the physics is lattice-based or continuous, 2D or 3D. I
hope this helps.